Side Finds

On three trading days — September 7, 11 and 14, 2026 — Gerresheimer AG reported several managers' transactions, all of them purchases by Active Ownership Opportunities SCS. On September 14 the voting rights notification followed: supervisory board member Klaus Röhrig now has 14.96 percent of voting rights attributed to him — 5,166,594 of a total 34,540,000 shares, up from 13.67 percent. The threshold was touched on September 11, 2026. Together with reported instruments of 3.18 percent — options maturing through December 2026 and June and December 2027 — the group reaches 18.14 percent. The chain runs through Active Ownership Corporation S.à r.l., Active Ownership Fund SICAV SIF SCS and AOC Gecko S.à r.l.

On the notification itself: under its item 2 it is a voluntary group notification, triggered by a subsidiary touching a threshold — not a filing forced by a newly crossed step. On the aggregated basis of Section 39 of the German Securities Trading Act the group was already above that level: the preceding notification shows a total of 16.85 percent. A mandatory offer to all remaining shareholders would in any case only be due at 30 percent (Section 35 in conjunction with Section 29(2) of the German Securities Acquisition and Takeover Act); the stake is far from that.

On materiality: a stake of 14.96 percent is far above the five percent threshold of the share count. And the constellation is unusual — an investor with just under 15 percent of the voting rights attributed to him — 18.14 percent including the options — who at the same time sits on the body that has to decide on the vacant chief executive role, on two sale processes and on a refinancing of more than EUR 2 billion. At the annual general meeting of September 1, 2026, Röhrig was re-elected as a shareholder representative; only 41.59 percent of share capital was represented there. What that means in terms of weight is worth a sober calculation — on the basis of September 11, 2026, because at the annual general meeting on September 1 the group still held only 13.67 percent: only the 14.96 percent carry votes, not the 3.18 percent from instruments — those are settled in cash. Measured against a turnout like that of September 1 (41.59 percent of share capital), that would be around 36 percent of the votes represented — not a simple majority, but the largest reported single shareholder, and a blocking minority for resolutions requiring a three-quarters majority.

In its July 29, 2026 prospectus, Reformation says it intends to file an initial registration statement on Form S-8 covering 11,761,807 shares — roughly 19.9 percent of the 59,076,208 shares outstanding after the offering. The S-8 was in fact filed on July 29, 2026. Such registration statements become effective automatically upon filing, and shares issued under them are freely sellable afterwards.

Two sources feed that pool. First, the legacy stack: as of June 27, 2026 there were 5,838,178 options outstanding at a weighted-average exercise price of $6.86, of which 5,584,765 were already exercisable. Second, the new 2026 Omnibus Incentive Plan, which reserved 5,316,858 shares and under which the board granted roughly 0.8 million options and roughly 2.7 million restricted stock units on July 29, 2026. The diluted share count used in the earnings statement was still only 52,948,297 in the second quarter of 2026 — that is the number to watch if you want to see how fast the pie gets cut into more slices.

After the July 31, 2026 IPO, Reformation looks like a normally tradable stock. For now it is not. The prospectus spells out what was actually free to trade after the offering: the 14,062,500 shares sold, plus another 229,546 shares on September 1, 2026 from the partial exercise of the underwriters' option. Everything else stays tied up: 45,013,708 shares, or 76.2 percent of the entire share count, are restricted securities under Rule 144 and are additionally covered by lock-up agreements with the underwriters for up to 180 days after the prospectus date of July 29, 2026 — which points to roughly January 25, 2027. J.P. Morgan and Morgan Stanley may release the lock-up in whole or in part at any time.

Why that matters to the price: on one side sit roughly 14.3 million freely tradable shares; on the other, private equity firm Permira with roughly 49.2 percent and the Aflalo Family Trust with roughly 19.8 percent. When the lock-up lapses, the tradable supply can multiply on paper. On top of that, shares representing roughly 75.6 percent of the company are covered by a registration rights agreement and can be registered for sale at any time. For anyone judging how this stock is priced, that is material: until late January 2027 the price is measuring a very thin supply.

The balance sheet at June 30, 2026 states both numbers inside a single parenthesis: 1,097,141,091 authorized Class A ordinary shares against 20,712,924 issued, of which 19,147,732 remain outstanding after buybacks. On paper the board can therefore issue 53 times the current Class A base without asking shareholders again — and for Class B shares carrying 50 votes each, the annual report says explicitly that the board may issue them "without further action by our shareholders".

That headroom is not dead letter. The weighted average share count rose from 5,888,507 (2023) through 8,049,592 (2024) to 19,858,117 (2025) — and that is after two share consolidations at 2-to-1 and 10-to-1 within ten months. Two shelf registration statements (Form F-3) have been on file with the SEC since July 2025, and the company drew on them repeatedly during 2025. For 2025 the annual report also discloses warrants over a weighted average of 2,348,554 shares that were excluded from diluted earnings per share only because they would have improved the result.

Through December 31, 2025 the Nano Labs balance sheet had no line called "Derivative assets". At June 30, 2026 it carries RMB 41.7 million ($6.1 million) — roughly 5.7 percent of total assets and more than four times the cash balance. The half-year report explains the origin in half a sentence: the company entered into "decumulator agreements" with third parties. A decumulator is a structured forward arrangement in which an investor sells an underlying asset in instalments over a fixed period and, in the adverse case, must deliver a multiple of the agreed quantity.

The 20-F for 2025 spells out the risk of that construction: such arrangements involve "leverage, margin requirements and other mechanisms that can magnify losses as well as gains", and adverse price moves may trigger margin calls, forced liquidations or the need to post additional collateral on short notice. In the first half of 2026 the position already cost RMB 2.8 million. Anyone treating the stock as a pure BNB proxy is therefore missing a second, leveraged layer — one whose size appears in no press release, only in the balance sheet.

On September 3, 2026 the hedge fund Solus Alternative Asset Management LP filed the seventh amendment to its Schedule 13D on Bristow. Reported position: 2,849,710 shares, or 9.61 percent, calculated on the 29,642 thousand shares outstanding as of July 31, 2026. In the sixth amendment, dated June 16, 2025, the position was 3,076,796 shares, or 10.69 percent. The attached schedule lists three open-market sales: 15,000 shares on September 1 at a weighted average of $45.85, 12,045 shares on September 2 at $45.68 and 41 shares on September 3 at $45.75.

What stands out is not the size but the identity. The Form 4 filed alongside identifies Christopher Pucillo as a director of Bristow; on the Schedule 13D/A the same Pucillo signs as managing member of the general partner of Solus. The original Schedule 13D dates from June 22, 2020, days after the merger of Era Group and Old Bristow.

At 9.61 percent Solus remains by far the largest reported single holder — for comparison, the South Dakota Investment Council reported 6.2 percent on April 10, 2026. Anyone holding the stock should watch the next amendment: a founding-era holder selling in tranches six years after entry is a standing supply overhang, and a director whose fund is selling is always information.

Bristow reported net income of $21.2 million for the second quarter of 2026. What barely registers sits in the income tax note of the Form 10-Q for the quarter ended June 30, 2026: on pre-tax income of $21.3 million the company recorded an income tax expense of $0.1 million — an effective tax rate of 0.5 percent. A year earlier the rate was 39.1 percent. Bristow explicitly points to "higher tax credit utilization in Nigeria".

The size of the effect matters. At last year's 39.1 percent rate, quarterly net income would have come in around $13.0 million instead of $21.2 million — a difference of roughly 39 percent of the reported profit. For the first half of 2026 the effective rate was 9.6 percent, against 34.1 percent a year earlier.

This is not an accusation but a question of durability: tax credits in a single country are finite and depend on local earnings. If the rate normalizes, reported earnings per share fall even if the operating business is unchanged. The next quarterly report rolls this number forward.

On May 8, 2026, Xos and Saudi Arabia's Aljomaih Automotive Co. signed the third restatement of their 2022 convertible note. The 8-K of May 13, 2026 describes the core change in one sentence: the conversion price was cut from $71.451 to $12.00 per share — a reduction of 83 percent. The company also gained the right to force conversion once the daily volume-weighted average price exceeds $16.00 for at least twenty out of thirty consecutive trading days.

The arithmetic: on remaining principal of $15.5 million as of June 30, 2026, full conversion would create roughly 1.29 million new shares — about 9 percent measured against the 14,217,852 shares outstanding (August 7, 2026). At the old conversion price it would have been roughly 217,000 shares, or 1.5 percent. The price of the extension was not interest. It was ownership.

The note amortizes in ten quarterly installments between November 11, 2025 and February 11, 2028: four of $1.5 million, four of $2.0 million and two of $3.0 million. The June 30, 2026 balance sheet shows $7.5 million of that as current — against $13.2 million of cash.

Buried in the June 30, 2026 quarterly report under "Item 5. Other Information" is a number that is large relative to the company. On June 30, 2026, Giordano Sordoni — Chief Operating Officer, co-founder and board member of Xos — entered into a Rule 10b5-1 trading arrangement covering up to 883,125 shares. Orders under it become effective "no earlier than September 29, 2026" and run through July 1, 2027.

For scale: 14,217,852 shares were outstanding as of August 7, 2026. The plan therefore covers roughly 6.2 percent of all shares — more than twice what Xos issued through its at-the-market program and its registered direct offering combined during the entire second quarter of 2026 (1,469,610 shares). Such a plan is legal, filed in advance and exists precisely to keep an executive from trading on inside information; on its own it says nothing about the company's prospects. It does say something about supply in the market: for a company whose entire market value on September 4, 2026 worked out to roughly $42.7 million, an announced sale of that size is a price factor in its own right.

The context: Sordoni already sold 49,370 shares at $4.0014 on August 20, 2026 (Form 4). Chief Executive Dakota Semler went the other way, terminating his December 2025 plan for up to 245,000 shares on May 26, 2026 without executing it.

In fiscal 2025 Photronics repurchased 5.0 million of its own shares for $97.4 million, at an average of $19.52 apiece. Every repurchased share was retired within the same fiscal year. The effect shows up in the share count: from 63.35 million shares (August 29, 2024) to 58.96 million (June 4, 2026) — each remaining share became roughly seven percent more valuable without the company having to earn a dollar more.

Since then the program has stood still. Through the first three quarters of fiscal 2026 (to August 2, 2026) the cash flow statement shows zero under common stock repurchases, against $97.4 million in the prior-year period. The permission is sitting there: as of May 3, 2026, $27.6 million of the authorization remained unused, and the filing offers only an intention — depending on market conditions, the company may use some or all of it.

The money has not vanished, it has been redirected. Purchases of property, plant and equipment rose from $130.9 million in fiscal 2024 to $188.1 million in fiscal 2025, up 43.7 percent; through nine months of fiscal 2026 they reached $130.4 million. On top of that, as of May 3, 2026 the company carried $211.3 million in outstanding capital commitments and accrued equipment purchases, of which $199.0 million is expected to be funded within twelve months. For a shareholder, that is the real capital allocation decision of this cycle: buybacks today versus plants in the United States and Korea tomorrow.

Photronics counts as debt-free — on August 2, 2026 the balance sheet carried all of $3.9 million in current and $4 thousand in long-term debt. One obligation does not appear there at all. It sits in the management discussion of the Form 10-Q for the quarter ended May 3, 2026, in the liquidity section, and concerns PDMCX, the Chinese joint venture in Xiamen in which Photronics holds 50.01 percent and the Japanese printing group Dai Nippon Printing (DNP) holds 49.99 percent.

Under certain circumstances DNP has the right to put its interest to Photronics — in other words, to sell its stake and require Photronics to buy. The price follows its ownership percentage of the venture's net book value, and closing takes place, per the agreement, within three business days of obtaining the required approvals. The same paragraph sizes the exposure: as of May 3, 2026 Photronics and DNP each held a net investment of roughly $177.1 million in the venture, up from $160.4 million on October 31, 2025. The number is growing, because the plant retains earnings and both partners keep investing.

For scale: $177.1 million equals roughly 14 percent of the $1,280.7 million of equity attributable to Photronics shareholders on August 2, 2026, and roughly a quarter of the entire cash pile. The report states explicitly that as of its issuance date DNP had not indicated any intention to exercise the right. That is precisely why this is a watch item rather than an alarm: the trigger does not sit with Photronics but with a partner in Tokyo — and the number that would come due is restated in every quarterly report.

In April 2026 ProQR entered an agreement with the U.S. company Ginkgo Bioworks to supply services supporting AI-enabled research. It did not pay in cash: ProQR issued 3,030,303 of its own ordinary shares at a reference price of $1.65. In return it received a contractual credit of $5.0 million to be applied against future services over a three-year term. Measured against equity of EUR 79.477 million as of June 30, 2026, that is roughly 5.4 percent.

The clause behind it is unusual: if the credit is not used within the contractual term, or on certain termination events, the unused amount is repayable — through the return of shares or, if those shares are no longer held, in cash. At the same time ProQR is under no obligation to draw the services at all and may terminate at any time without penalty. In accounting terms the transaction was recognized solely within equity under IFRS 2; it becomes an expense only as services are actually received. For shareholders that means the dilution has already happened while the value received has not yet been consumed — and in the adverse case it turns into a cash liability. On July 15, 2026 ProQR registered the resale of precisely those 3,030,303 shares for the provider.

ProQR's balance sheet as of June 30, 2026 carries a current borrowing of EUR 5.017 million — 6.3 percent of total equity of EUR 79.477 million. It stems from a Dutch government innovation credit awarded in December 2018 for the eye medicine sepofarsen; EUR 3.907 million was drawn between 2018 and 2022. ProQR repaid EUR 1.008 million in the fourth quarter of 2023, leaving EUR 2.899 million of principal plus EUR 2.118 million of accrued interest.

The unusual part sits in Note 9 of the interim report: repayment of the full EUR 4.292 million including interest could be waived if conditions are met, subject to annual review. In December 2025 the waiver was extended once more, in the filing's own words, "until December 31, 2026." For shareholders that is a two-sided coin with a fixed calendar date. If the waiver becomes permanent, a one-off gain in the order of EUR 4.3 million appears — roughly 19 percent of the EUR 22.111 million half-year loss. If it is not extended, the amount falls due. The sepofarsen program itself was discontinued in August 2022 and the rights sold to Théa in December 2023; the loan behind it is still on the books.

In the second quarter of 2026 Ring drilled three horizontal wells of roughly two miles in Crane County, Texas, each with a 100 percent working interest. The Form 10-Q for the quarter ended June 30, 2026 adds a sentence that is easy to skim past: these are the first laterals of that length the company has drilled in an area historically developed with vertical wells. They were not completed until the third quarter of 2026, so no production figures exist yet.

More than three wells hang on the outcome. In the same week Ring published its first guidance for 2027: 20 to 30 new wells longer than 1.5 miles and capital spending of $135 million to $165 million. That equals 34 to 42 percent of the $396 million market capitalization (as of September 8, 2026). Whether the program delivers will first be tested on those three two-mile wells. The first hard evidence lands in the report for the quarter ended September 30, 2026, which will show whether output falls inside the 13,000 to 13,950 barrels of oil per day that Ring has guided for the second half of 2026.

Note 1 of the Form 10-Q for the quarter ended June 30, 2026 carries a paragraph headed “Correction of an Immaterial Error” that appeared in no press release. While preparing the first quarter 2026 statements, the company found that certain revenues held in suspense from fiscal years 2017 through 2025 had been assigned incorrectly. The stated cause: ownership interests of other interest owners had been set up wrongly from the start, without accounting for Ring's recovery of capital and normal operating costs.

Prior periods were revised. Accounts payable as of December 31, 2025 fell by $7.3 million, retained earnings as of December 31, 2024 and 2025 rose by $5.7 million each, and deferred income taxes rose by $1.5 million. The company explicitly considers the error immaterial, and auditor Grant Thornton had issued an unqualified opinion on internal control over financial reporting as of December 31, 2025. The timing is what makes it interesting: new chief financial officer Sundip “Sonu” S. Johl started on February 27, 2026 — the correction surfaced in the first quarterly close under his responsibility. Whether it stays at one correction will show in the next audit opinion.

The “Condensed Operating Data” table in the second quarter 2026 earnings release contains a line worth reading twice. The average realized price for natural gas is minus $5.20 per thousand cubic feet (Q1 2026: minus $2.54; Q2 2025: minus $1.31). A negative sales price means exactly what it says: Ring pays the buyer to take the gas. The cause is geography, not management — in the Permian Basin natural gas comes up as a by-product of oil production, takeaway capacity is tight, and the regional benchmark regularly slips below zero.

The scale is not trivial. Ring sold 1,764,659 thousand cubic feet of natural gas in the second quarter of 2026. Multiplied by minus $5.20 that is a drag of roughly $9.2 million in a quarter with $104.7 million of total revenue and $24.0 million of company-adjusted net income. Ring hedges part of the regional spread — as of June 30, 2026 with Waha basis swaps covering 374,623 thousand cubic feet for the third quarter of 2026 at an average spread of $2.15. Anyone tracking the number will find it in the same line of every quarterly report.

On June 16, 2026 Lakeland filed a registration statement on Form S-8 covering 700,000 shares for the new 2026 equity incentive plan, which shareholders had approved at the annual meeting the same day. Measured against the 9,869,164 shares outstanding on June 5, 2026, that is 7.1 percent — your share of the company can shrink by that fraction without a dollar coming in from outside.

It does not stand alone. The weighted average share count had already risen from 7.43 million in fiscal 2025 to 9.63 million in fiscal 2026 (up 30 percent), largely through the January 24, 2025 equity raise of 2,093,000 shares at $22.00. Stock-based compensation expense more than doubled in fiscal 2026 to $3.4 million, and $5.8 million remained unrecognized as of January 31, 2026. The share repurchase program, with $5.0 million still authorized, went unused.

On March 27, 2026 Lakeland sold the inventory and intellectual property of its High Performance and High Visibility product lines to National Safety Apparel, LLC for $14.0 million, of which $13.2 million came in as cash. The resulting $6.5 million disposal gain is the sole reason the first quarter of fiscal 2027 closed with $0.4 million of net income; without it the operating result would have been minus $4.2 million.

The quiet arithmetic sits in the annual report: those same two lines generated $13.2 million of revenue in fiscal 2026 (High Performance Wear $8.1 million, High Visibility $5.1 million) — 6.9 percent of group sales of $192.6 million. Into precisely that hole, the chief executive reaffirmed on June 9, 2026 a target of high single-digit revenue growth for fiscal 2027. First-quarter revenue grew 1.4 percent.

Note 6 of the quarterly report for the period ended April 30, 2026 carries a sentence that appeared in no press release: on April 13, 2026 Bank of America entered into a limited waiver releasing Lakeland from its non-compliance with two financial covenants as of January 31, 2026. Three days later the company filed its annual report. The $40.0 million commitment, the December 12, 2029 maturity and the interest rate were unchanged; as of April 30, 2026 Lakeland states it was back in compliance with all covenants.

The point is the schedule that follows. The loan agreement requires a basic fixed charge coverage ratio of at least 1.20x and a funded debt to EBITDA ratio of no more than 3.5x — stepping down to 3.25x on February 1, 2026 and to 3.0x on February 1, 2027. The limit therefore tightens while the fiscal 2026 operating result stood at minus $15.5 million and operating cash flow was negative by roughly $16 million in each of two consecutive years. Of the facility, $23.8 million was drawn and $16.2 million available on April 30, 2026.

The March 17, 2023 SPAC business combination left 15,666,526 warrants outstanding, each exercisable for 1.11829212 NioCorp common shares at $11.50 per bundle and expiring on March 17, 2028. They are by far the largest block within the 19,101,667 warrants outstanding at March 31, 2026 and pull the weighted average exercise price up to $9.78.

In practice this is a switch with two sides. If it is flipped — that is, if the share price holds above $11.50 — these warrants would deliver roughly $180 million to the company, a quarter of the roughly $722 million funding gap, without a single loan agreement. In exchange, up to 17.5 million new shares would be created, a little over twelve percent of the March 31, 2026 count. The company states it does not expect exercises unless the market price reaches or exceeds the relevant exercise price. Anyone tracking the financing question should know this line item: it is the only sizeable source of capital that depends neither on EXIM nor on another placement.

The 2022 feasibility study that carries the entire economics of the Elk Creek project projects gross revenue of $21.9 billion over 38 years. Working through its own tonnage and price assumptions, roughly $13.5 billion comes from scandium oxide (3,676 tonnes at $3,674 per kilogram), $8.0 billion from niobium and only $0.4 billion from titanium dioxide. Scandium therefore carries about 62 percent of the projected life-of-mine revenue — for a metal traded in tiny volumes compared with niobium and titanium.

The problem appears in the same annual report (10-K) for fiscal 2025, two sections earlier: the commodity price table there lists a U.S. price for scandium oxide of $1,200 per kilogram for 2024, after $2,100 (2022), $2,200 (2021) and $3,800 (2020), sourced to the USGS. The study assumption is therefore roughly three times the most recent market price disclosed in a filing. The company itself cautions that these pricing surveys may not be representative of what it would actually realize — a caveat that cuts both ways. NioCorp is working on an updated feasibility study; the scandium price it adopts will decide the single largest line item in the entire project calculation.

Aura Biosciences raised $280.8 million net in May 2026 and puts its cash runway at the first half of 2029 in the Form 10-Q for the quarter ended June 30, 2026. Anyone concluding the funding question is settled for years misses two decisions taken in a single week of August. On August 5, 2026 shareholders approved a charter amendment raising authorized share capital from 150,000,000 to 500,000,000 shares, against 103,704,809 shares actually outstanding on August 6, 2026. On August 11, 2026, the day of the quarterly report, the company filed a new shelf registration on Form S-3 for $500.0 million, including a $200.0 million at-the-market program that can sell shares gradually into the open market (sales agent Jefferies, 3.0 percent commission).

Measured against a market capitalization of roughly $784 million (price $7.56 on September 4, 2026), that at-the-market program alone equals about a quarter of the entire company. This is not an accusation — a shelf is authorization, not a sale, and for a company whose pivotal data are not due before the second half of 2027, preparation makes sense. But the direction is unmistakable: the share count already climbed from 64,150,468 (March 24, 2026) to 103,704,809 (August 6, 2026), and the frame for the next round is already open.

The global settlement over the Maui wildfires is treated as done: the last condition was satisfied on April 10, 2026, the first installment was paid, the lawsuits are being dismissed. One line in Note 2 of the Form 10-Q for June 30, 2026 shows that a remainder is still open: "Approximately 80 plaintiffs who 'opted out' of the class Settlement Agreement have not signed individual agreements and releases to join the global settlement." Some of those opt-outs are pursuing their claims in state and federal court, including two putative class actions.

For those cases the defendants have reserved $500 million out of the settlement total — roughly a quarter of the $1.92 billion market capitalization (as of September 8, 2026). The company writes that, based on current projections, it believes the holdback will be sufficient. That sentence is the watch point: if the money falls short, the excess lands on a balance sheet that already faces three installments of $479 million. The status is updated in Note 2 of every quarterly report.

In September 2024 Hawaiian Electric Industries registered an at-the-market program with the U.S. securities regulator, the SEC: permission to sell, at its sole discretion and without further announcement, up to $250 million of its own common stock into the market. The Form 10-Q for June 30, 2026 records one easily missed sentence about it: "To date, HEI has not sold any common stock under this program." Not a single dollar had been drawn. The program equals roughly 13 percent of the $1.92 billion market capitalization (as of September 8, 2026) — and the filing counts it explicitly as part of the $1,339 million of "available liquidity."

The context is what makes the find price-relevant: three settlement installments of $479 million each remain payable through April 2029, and the company says it is working with financial advisors on a financing plan. Anyone who wants to see the moment the option becomes a fact does not need a press release — the amount drawn appears in the same liquidity table of every quarterly report, and the share count sits on the cover page (most recently 172,728,004 shares as of July 31, 2026).

Buried in the risk factors of the 2025 annual report (10-K) is a sentence that is easy to miss in launch euphoria. Kura writes that initial sales of KOMZIFTI may deplete the “prevalence pool” of patients in the approved indication more quickly than expected, with a negative impact on future sales. The logic: in a rare disease there are two patient groups. Prevalence is the standing population of everyone living with the condition today; incidence is the number of new cases each year. A newly approved drug harvests the standing population first — after that, only the new cases remain.

For Kura this is measurable. In the second quarter of 2026 the company reported roughly 115 new patient starts (35 percent more than in the first quarter) and more than 250 total prescriptions. As long as new starts keep rising, the standing population has not been worked through. If that number rolls over while total prescriptions keep climbing, the warning in the company’s own report has arrived — and the growth case behind a market capitalization of roughly $1.18 billion (share price $13.25 on September 4, 2026) shifts entirely into the frontline setting, which is not yet approved.

Three board seats were up for election at the annual meeting on June 4, 2026. Two results stand out. Mary T. Szela, chair of the compensation committee, received 38,859,502 votes for and 19,821,530 withheld — one in three votes cast declined to support her. Diane Parks, also a member of the compensation committee, drew 44,060,724 for and 14,620,308 withheld. The third nominee, Michael J. Vasconcelles, received 58,169,894 for and only 511,138 withheld. Those 19.8 million withheld votes equal roughly 22 percent of all 88.95 million shares outstanding (as of June 30, 2026) — not a formality, but a protest.

Seven weeks later, on July 24 and 27, 2026, the company entered into new employment agreements with its three most senior executives. The substance: enhanced severance in the event of a Corporate Transaction, meaning a takeover. Chief Executive Troy E. Wilson would then receive 24 months of base salary in cash, 200 percent of his target bonus, 24 months of health coverage and full acceleration of every outstanding equity award; Chief Commercial Officer Brian Powl and Chief Operating Officer Kathleen Ford would each receive 18 months of salary and 150 percent of target bonus. A company that sharpens those clauses for three executives at once is not planning for nothing — and the full agreements will only be filed with the third-quarter 2026 report.

On September 8, 2026 a fourth agreement of the same design followed: with Jennifer Fulk, Kura appointed a dedicated Chief Financial Officer for the first time since February 2022 — after Marc Grasso stepped down as CFO effective February 4, 2022, Chief Executive Wilson had carried the principal financial officer role himself. Her agreement, too, steps severance up on a “Corporate Transaction” closing within 18 months (Form 8-K filed September 8, 2026, Item 5.02).

The notes to the 10-K for 2025 contain a policy you would not expect at a software company: "The Company provides software licenses as barter transactions in exchange for other assets, such as trade credits, in the ordinary course of business. Trade credits may be redeemed for advertising media, VDR revenue share or other goods and services made available by certain partners …" In plain English: Veritone hands over software licenses and receives no money but credits — redeemable for advertising inventory, for a share of data-business revenue, or for other goods and services from partners. The barter is measured at the estimated fair value of what is received; where that is not reliably estimable, at the standalone selling price of the software. The proceeds then run straight through the revenue line.

The size of it shows up in the cash flow statement, because it has to be backed out as a non-cash item: $1.511 million in fiscal 2025 and $0.849 million in the first half of 2026. And that valuation is precisely where the third quarter of 2025 broke. According to the 8-K of April 14, 2026 (Item 4.02), management had misvalued the consideration received for an on-premise software delivery — $2.2 million of overstated revenue, "approximately 8%" of quarterly revenue. The resulting material weakness relates explicitly to "non-routine revenue transactions" and had not been remediated as of December 31, 2025.

For the fiscal year ended May 31, 2026 Southern Cross Gold reports a net loss of C$5,475,812 — for a company worth roughly C$3.25 billion that sounds like almost nothing. A few pages later the MD&A carries the other number: C$32.1 million of net expenditures over the same period. The gap is not a contradiction but an accounting rule: drilling and decline development are not expensed, they are capitalized as an asset. As at May 31, 2026 that produced C$119,617,908 of capitalized exploration costs on the balance sheet — practically the same amount as the entire cash position.

Capitalizing is permitted and standard in the industry. But it means the income statement of an explorer says very little about how fast the money goes out — and that the largest single balance sheet item holds its value only for as long as nobody has to write it down. Auditor D&H Group LLP named exactly that question a key audit matter in its report of August 25, 2026: the assessment of whether indicators of impairment exist for the exploration and evaluation assets. For fiscal 2026 management found none.

In the third quarter of fiscal 2026 Southern Cross Gold filed a C$700 million base shelf prospectus with the securities regulators of British Columbia, Alberta and Ontario (release of April 8, 2026). A shelf prospectus is not a sale; it is pre-approval held in reserve. It lets the company issue shares, warrants, debt securities, subscription receipts or units at any point over 25 months without going through the full approval process again. The MD&A as at May 31, 2026 states that as of the report date nothing had been issued under it and that the company has no immediate plans — "and may never issue any securities under the base shelf prospectus."

For scale: C$700 million is roughly one fifth of the C$3.25 billion market value as of September 4, 2026. By comparison, the company's only large financing to date raised gross proceeds of C$143.1 million in fiscal 2025, at C$4.50 per share. The prospectus says nothing about whether or when it will be drawn. It does say the machinery is in place — which is exactly why the "Financings" section of the next quarterly report is worth reading.

At the annual general meeting on December 8, 2025, Frontline shareholders approved a twelve-month authorization: the board may place up to 377,377,111 new shares — equal to 169% of the 222,622,889 shares outstanding as of June 30, 2026 — without pre-emption rights for existing shareholders, at a minimum price of $1.00 per share. This is a common financing authorization among shipowners for future fleet purchases, but its size exceeds the 5% materiality threshold many times over.

None of the authorization has been used so far: the share count remained unchanged at 222,622,889 shares between year-end 2025 and June 30, 2026 per the statement of changes in equity in the interim report. The authorization expires on December 8, 2026 — until then, the board could in theory place new shares at any time without a shareholder vote.

In January 2026, Frontline agreed to buy nine new VLCC tankers for a total of $1,224.0 million — from shipyard contracts the earnings release attributes to "affiliates of Hemen." Hemen Holding Limited is Frontline's largest shareholder with 35.6%, held per the annual report (Form 20-F for 2025) through two family trusts of shipowner John Fredriksen. As of June 30, 2026, $601.1 million of this deal remained unpaid.

The transaction size equals roughly 21% of total assets as of June 30, 2026 ($5,812.7 million) — well above the materiality threshold. The annual report's Note 20, "Related Party Transactions," lists the ongoing business with Hemen-affiliated companies (a combined $9.8 million of revenue and $4.1 million of expenses in 2025), but does not mention an independent fairness review or a special committee for the $1.224 billion newbuilding contract itself — it was agreed only in 2026 and will therefore first appear as a related-party item in the next annual report.

The 2025 impairment charge wrote off ₩77.9 billion of goodwill, largely tied to the GC Cell business per press coverage. In the same fiscal year, ₩70.3 billion of brand-new goodwill arrived from the ABO Holdings first-time consolidation — leaving the balance-sheet goodwill figure nearly unchanged (₩125.6 billion to ₩118.1 billion) even after a headline write-down. The new goodwill's value depends on the U.S. plasma-collection business ABO represents — the same business whose operating profit collapsed 93.8 percent in Q2 2026.

If U.S. plasma weakness persists into year-end, that ₩70.3 billion of relatively fresh goodwill is a plausible candidate for the next impairment test.

Beyond the ₩308.7 billion already paid, the Curevo divestment filing discloses a further contingent milestone payment of ₩153.31 billion ($101.3 million) due if a sales target is met, with an expected final date of Apr 15, 2035. That contingent amount alone equals roughly 11 percent of GC Biopharma's current market capitalization — and because it depends on a future sales threshold for Eli Lilly's shingles vaccine, it appears nowhere on GC Biopharma's balance sheet today.

The trigger sits entirely outside GC Biopharma's own filings: it depends on Eli Lilly's regulatory and commercial progress with the shingles vaccine (branded amezosvatein) that came with the Curevo deal.

GC Biopharma's Jul 9, 2026 divestment filing shows Curevo carried on the books at just ₩43.2 billion, against an upfront payment from Eli Lilly of ₩308.7 billion ($204.0 million) that closed Jul 8, 2026. The upfront payment alone equals more than 22 percent of GC Biopharma's entire market capitalization (₩1,393.5 billion, Aug 21, 2026). Analyst consensus already prices this in: Q3 2026 EPS of ₩20,358.76 towers over the ₩4,190 consensus for full-year 2027, a gap that only makes sense if the 2026 figure is overwhelmingly a one-off gain rather than a step-change in earning power.

Whether the market treats this correctly — pricing the gain as non-recurring rather than extrapolating it — becomes visible the moment the Q3 2026 report lands and analysts either strip the gain out of forward estimates or don't.

As of June 30, 2026, BigBear.ai carries $238.6 million of goodwill — 28 percent of its total assets of $841.3 million. Most of it (provisionally $192.7 million, revised to $190.1 million in the 10-Q for the quarter ended June 30, 2026) arose on December 31, 2025, when the company bought Ask Sage for $271.6 million in cash. Goodwill is the premium paid above the tangible value of what was acquired: essentially the price of a brand, a customer base and an expectation. It does not wear out over time; it only disappears through an impairment charge once the expectation no longer holds.

In the same footnote, the 10-K for 2025 states that no safety margin remains for that goodwill: "the fair value of the reporting unit approximates its carrying value," and an adverse change in key assumptions may result in a future impairment that could be material. At this company that is not a theoretical warning: BigBear.ai has already written goodwill down three times — $53.5 million (2022), $85.0 million (2024) and $70.6 million (2025), a combined $209.2 million, more than its entire 2025 revenue. On top of that came a $53.4 million long-lived asset impairment in 2025. The annual impairment test is performed as of October 1 per the 10-K; its outcome shows up in the next annual report.

In August 2020 Zalando issued two tranches of unsecured convertible bonds of EUR 500 million each. Tranche A ran five years at a 0.050 percent coupon and was repaid in 2025 with EUR 400.0 million, after EUR 100.0 million of principal had already been repurchased in 2024 for EUR 95.5 million. Tranche B runs seven years, carries a 0.625 percent coupon and therefore matures in August 2027. As of June 30, 2026 it sat in non-current liabilities at a carrying amount of EUR 487.0 million, with a fair value of EUR 480.0 million.

The conversion price is what matters. It was set at issue at EUR 92.25 — 50 percent above the then reference price of EUR 61.50. At the XETRA close of EUR 22.78 on August 21, 2026 the bond is nowhere near conversion; what was designed in 2020 as a possible equity building block is today a plain cash repayment obligation. EUR 487.0 million equals roughly 35 percent of the EUR 1,397.8 million cash balance and roughly 19 percent of equity. With free cash flow at minus EUR 95.0 million in the first half of 2026, whether Zalando refinances or pays out of cash is not an academic question — and the answer is a data point on this group's financial discipline.

Anders Holch Povlsen, chief executive of the Danish fashion group Bestseller, has sat on Zalando's supervisory board since December 2013 and held 10.1 percent of the shares as of December 31, 2025 according to the annual report, making him the largest individual shareholder. The notes to the Half-Year Report 2026 show what that means operationally: in the first half of 2026 Zalando ordered EUR 184.8 million of merchandise from related parties, against EUR 122.4 million in the prior-year period — up roughly 51 percent. At the reporting date this left EUR 152.6 million of liabilities (December 31, 2025: EUR 225.5 million), of which EUR 103.3 million was owed to a reverse factoring provider. Measured against equity of EUR 2,565.8 million, the order volume equals a good seven percent and the liabilities close to six percent. Zalando states that all transactions were conducted on an arm's length basis and discloses them.

A second role was added in 2025: Zalando acquired 38,740,244 ABOUT YOU shares for EUR 251.8 million from Aktieselskabet af 12.6.2018, a company controlled by that supervisory board member. This disclosure was missing from the notes to the 2025 consolidated accounts — BaFin formally established that as an error on July 21, 2026 without imposing a fine, and Zalando added it in the Half-Year Report 2026 invoking IAS 8.41. For investors the setup itself is the thing to watch: a supplier of this size who is simultaneously overseer and largest owner earns a second look every time purchasing terms, gross margin or further related-party transactions come up.

The interim group management report as of June 30, 2026 carries a number that appears in no earnings release: EUR 654.5 million of supplier claims against Zalando had been transferred to factoring providers at the reporting date — after EUR 931.6 million on December 31, 2025 and EUR 639.2 million on December 31, 2024. Under this reverse factoring a supplier sells its claim to a bank and is paid immediately, while Zalando pays later. The Annual Report 2025 quantifies the difference: payment terms under reverse factoring run 60 to 180 days, against typically 45 to 90 days for comparable suppliers outside those arrangements. The balances are reported not as financial debt but under trade payables, so payments to the factoring providers flow through operating cash flow.

The EUR 277.1 million decline equals roughly a tenth of the EUR 2,565.8 million equity base, and the balance itself roughly a quarter. The effect is directly visible in the accounts: the line "increase/decrease in trade payables and similar liabilities" reads minus EUR 234.4 million for the first half of 2026, against plus EUR 26.3 million a year earlier. Operating cash flow fell from EUR 140.8 million to EUR 35.1 million and free cash flow from plus EUR 17.2 million to minus EUR 95.0 million. For investors this is the lever behind the very metric that puts Zalando into three value rankings: stretch your payment terms and you manufacture cash flow without selling one extra item — pull them back in and you lose it just as fast.

The Helus Pharma earnings release for the quarter ended June 30, 2026, gives cash, loss and trial progress. What it does not give sits in the interim financial statements of the same filing, under “Subsequent Events”, note 13: “On July 1, 2026, the Company issued 7,889,846 Common Shares upon vesting of RSUs.” In a single day, the vesting of restricted share units created 7.89 million new shares — roughly 10.8 percent of the 73,060,172 shares outstanding as of the date of that report. Of the 4,103,974 restricted share units granted the same day, 4,033,304 vested immediately.

Those shares are not immediately free: they sit under lock-up agreements dated July 1, 2026. But the lock is already loosening. The filing states that acceleration provisions in some of those agreements were triggered on August 13, 2026, releasing the restrictions on 1,916,362 shares early; 2,116,942 shares remain subject to lock-up. For a stock that ran from $8.06 to $13.99 between August 3 and August 25, 2026, that is a hard number: a double-digit percentage of the capital sits with people who received it at no cost, and part of it has been sellable since August 13, 2026.

The balance sheet of crypto exchange Gemini carries a line as of June 30, 2026 that is rarely this large at a listed company: $258.8 million of related party loans. The related party is Winklevoss Capital Fund, LLC (WCF), the family office of co-founders and controlling shareholders Cameron and Tyler Winklevoss. What was borrowed is not cash but bitcoin: 4,419 bitcoin were outstanding at the reporting date, at fees between 4.0 and 5.0 percent per year. For comparison, unrestricted cash and cash equivalents on the same date were $188.6 million, and reported stockholders' equity was $468.3 million.

The decisive sentence sits in Note 14 of the quarterly report: the loans have no stated maturity date but are callable upon written notice by WCF — and Gemini then has until the end of that same business day to return everything outstanding. Part of that borrowed bitcoin is simultaneously pledged as collateral to lender Galaxy Digital ($75.0 million outstanding as of June 30, 2026). Economically the interests are aligned; the founders are unlikely to push their own company into a squeeze. Structurally it remains a callable dependency on a single counterparty that also holds roughly 94.7 percent of the voting power. Anyone tracking the stock should read the related party loans line in every new quarterly report.

Buried in the "Commitments and contingencies" note of the Form 10-Q for the period ended June 30, 2026 sits a matter that has nothing to do with bitcoin. On April 21, 2020 the predecessor company — then the cannabis software vendor Akerna — took out a $2.2 million loan under the U.S. Paycheck Protection Program. On September 3, 2021 the Small Business Administration forgave it in full. On January 25, 2024 the same agency wrote that it was reconsidering: as a software provider to the cannabis industry, Akerna may never have been eligible. The Department of Justice issued a civil investigative demand in parallel.

In July 2026 American Bitcoin settled the matter for approximately $2.5 million, with no admission of liability. Against total assets of $1.31 billion the amount is small. Measured against cash of $18.4 million on June 30, 2026 it is roughly 14 percent, and against the quarter's general and administrative expenses of $7.7 million it is about one third. What stands out is the length of the chain: a 2018 blank-check company became cannabis software, the software company became bitcoin miner Gryphon, the miner became American Bitcoin — and the bill from the first life arrived six years later.

The 10-Q for the quarter ended June 27, 2026 contains a sentence with an expiration date: the board extended its authorization to repurchase up to 80 million shares (split-adjusted) only through July 2026. Since the first authorization in 1999, 39.2 million shares have been bought back under it — nearly half the available room. In 2025 alone, $243.6 million went into buybacks, plus another $76.4 million in the first half of 2026.

That sets up a small but trackable decision in the weeks ahead: if the board renews it, the second payout channel alongside the dividend stays open — with $1.42 billion in cash, the company could easily fund it. If it doesn't renew, that would suggest the cash is earmarked elsewhere, for example further acquisitions following Bison ($138.3 million, March 2026) and Chicago Extruded Metals ($3.9 million, June 2026). Either is defensible — but they are two different signals, and both will show up in the next quarterly report.

Buried in the market-risk disclosures of the 2025 annual report is a line easy to skim past. As of December 27, 2025, Mueller Industries held open forward contracts to sell roughly $164.9 million of copper over the following twelve months — and to buy back only roughly $16.6 million. On the futures market, the company is net positioned for roughly $148 million against a falling copper price.

This isn't speculation — it's the flip side of the inventory position: Mueller is hedging the value of its copper stock against a price drop; the report itself says the contracts manage "price risk associated with inventory." Economically, though, it means part of what a further copper rally would add to inventory value has already been sold off. Anyone buying the stock as a bet on copper is buying a hedged bet. That's good for earnings, bad for upside imagination.

Tripod's corporate website lists three business divisions: printed circuit boards, industrial automation and an "innovation" division. The segment note in the 2025 annual report tells a different story. The PCB business contributed NT$73,131 million in revenue — 99.6% of consolidated revenue — and the entire segment operating profit of NT$13,646 million (consolidated operating profit after unallocated corporate costs: NT$12,912 million). The remainder, grouped as "other", contributed NT$268 million in revenue and a loss of NT$89 million.

That is not a red flag on its own — at 0.4% of revenue the segment cannot endanger the balance sheet — but it frames the equity story: buying Tripod shares means buying almost exclusively the PCB business. The additional divisions remain a footnote with no discernible growth contribution so far.

Tripod's investor presentation contains — updated monthly — a ranking of the world's 20 largest PCB makers by revenue, sourced from market researcher Prismark. Comparing the November 2023 and August 2026 editions shows a steady slide: Tripod ranked 6th in 2021 ($2,257 million revenue), 7th in 2022 ($2,218 million, -1.7%), still 7th in 2023 ($1,919 million), 8th in 2024 ($2,050 million) and 9th in 2025 ($2,361 million, +15.2%). Tripod's own revenue growth is real, and strong in the most recent year — but slower than the most aggressive Chinese competitors: Shennan Circuits grew 32.2% in 2025, Wus Group 40.8%, Victory Giant 79.9%.

The August 2026 deck also supplies the structural backdrop: China's share of global PCB output rose, per Prismark data, from 53.7% (2019) to a projected 57.7% (2026) — at the expense of Taiwan, Japan, Korea and every other location combined. Tripod is riding the AI boom, but it is losing relative ground.

Note 27 of the 2025 annual report (financial risk management) discloses just how concentrated Tripod's credit risk is: 52% of all trade receivables at 31 December 2025 were owed by the company's ten largest customers — down from 56% a year earlier. The report does not name individual customers, which is standard practice in Taiwanese annual reports, but the scale is unmistakable: a single large customer shifting orders or switching supplier would hit Tripod disproportionately hard.

The direction is at least encouraging — the share fell from 56% to 52%, suggesting the extra demand from the server and memory boom is broadening rather than narrowing the customer base. Still, it is a metric worth watching, especially because server and memory customers are among the most concentrated buyer groups in the entire electronics industry.

Between December 31, 2025 and June 30, 2026, Iljin Electric's short-term borrowings rose from KRW 72.3 billion to KRW 143.2 billion per the semi-annual report — a doubling in six months. Over the same period, trade receivables grew from KRW 389.3 billion to KRW 455.1 billion and inventories from KRW 335.5 billion to KRW 386.6 billion, a combined increase of KRW 116.9 billion. The group's debt-to-equity ratio climbed from 159.2 percent (end of 2025) to 161.1 percent (06/30/2026) as a result.

That's not an alarm bell by itself — cash (KRW 137.6bn to KRW 188.6bn) and equity (KRW 588.7bn to KRW 654.2bn) grew over the same period too, and rising working capital is normal in an order boom. Still, the pace stands out: when receivables and inventories grow faster than operating cash flow, the gap has to come from borrowed money — and the company's A- credit rating hasn't been updated since May 2024, even though revenue and balance sheet structure have shifted markedly since.

The semi-annual report filed August 11, 2026 documents that parent company Iljin Holdings cut its stake from 49.25 to 42.98 percent between January 1 and June 30, 2026 — from 23,483,712 to 20,496,998 shares, a drop of 2,986,714 shares. The mechanism: the full exercise of an exchangeable bond issued October 20, 2025 (1,895,195 shares, of which 710,698 on February 26 and 1,184,497 on April 28, 2026), plus a price return swap contract announced April 17, 2026, covering 1,160,093 shares, of which 1,091,519 had settled by May 20, 2026.

What stands out is the timing relative to the price high: the monthly average price rose from KRW 62,367 in January to KRW 121,022 in May 2026 (intraday high KRW 144,100 on May 4), and most of the reported transactions fall exactly inside that window. Both instruments are legally clean, publicly disclosed transactions — the report does not state the motives. Anyone tracking the stock should check the next report to see whether the stake keeps shrinking or stabilizes around 43 percent.

The 2026 half-year report is explicit: „당기와 전기 중 단일 외부 고객으로부터의 수익이 연결기업 총 수익의 10% 이상인 주요 고객은 없습니다.“ — no single customer accounts for 10 percent of group revenue. That is a genuine strength compared with many suppliers. But anyone concluding from that alone that Cosmecca is a direct beneficiary of the K-Beauty export wave is off: direct exports from Korea made up just 11.4 percent of first-half 2026 revenue (47.0 of 411.2 billion won) — a share that has held steady between 10.6 and 12.1 percent since 2023.

Overseas growth instead runs almost entirely through consolidated foreign subsidiaries that produce and sell locally — chiefly Englewood Lab in the US. That makes no difference to the revenue story, but it matters for understanding the company: structurally, Cosmecca is less an exporter than the operator of three separate country entities.

While US subsidiary Englewood Lab nearly doubled operating profit in 2025 (37.4 billion won, up 98.8 percent), the China entity reported an operating loss of 1.9 billion won — the second straight loss year by the company's own account, on revenue that fell 15.9 percent to 34.0 billion won. The annual report cites the relocation of the R&D unit to Shanghai and a shift toward locally developed products as the response.

On its own, that is not an alarm bell — the China business is small at roughly 5 percent of group revenue. But it shows that the "K-Beauty boom" does not lift every market equally: while Korea and the US grow, China has shrunk at Cosmecca for two years running.

The 2025 annual report spells out the mechanism in one sentence: „자본잉여금은 전년 69,037백만원에서 62,803백만원으로 6,234백만원(9.0%) 감소하였습니다. 이는 잉글우드랩㈜ 지분 추가 취득(공개매수 방식, 지분율 39% → 50%)에 따른 종속기업 지분 변동 효과가 자본잉여금에서 차감 반영된 결과입니다.“ — translated: capital reserves fell by 6,234 million won because the tender-offer buy-up of US subsidiary Englewood Lab from 39 to 50 percent was booked directly against equity, not through the income statement. In the first half of 2026, a second round followed: a tender offer for up to 66.7 percent, worth roughly 43 billion won.

Result: despite a 39.9 billion won first-half profit, reported group equity fell from 336.2 billion won (December 31, 2025) to 334.6 billion won (June 30, 2026) — an effect that never shows up in the income statement. The next annual report will show whether and how strongly this effect continues into 2026.

Digital Daesung extended a short-term loan to subsidiary Igam that grew, per the fiscal 2025 annual report, from KRW 18.0 billion to KRW 23.0 billion over the year — purpose stated explicitly as "funds to acquire shares of another company." In other words: Igam receives money from its own parent to go buy yet another company — in the same year its own operating profit came in 47.6% below the purchase-time forecast and its second straight year of shrinking results.

Who is being bought, and at what price? The annual report names neither a target nor terms — only the credit line itself. For assessing capital discipline within the group, that matters: capital does not just flow from the parent into a shrinking subsidiary, it flows onward from there into a third, so-far unnamed target.

On May 29, 2026, Digital Daesung ran an unusual same-day double transaction: the group handed over 551,204 treasury shares at KRW 8,050 — the prior day's closing price — to two Igam executives, Kwon Jong-cheol and Cho Seong-jin, 275,602 shares each, worth KRW 4.44 billion combined. The same day, Digital Daesung bought 800,000 Igam shares (9.9%) for KRW 8.87 billion at KRW 11,093 per share — paid partly in cash, partly through this very share transfer.

The treasury shares do not go to the open market; they go directly into the hands of managers at the subsidiary who are simultaneously the sellers of the Igam stake — a "paper for paper" swap between two closely linked parties. For Digital Daesung shareholders, that means 551,204 previously dormant treasury shares become economically active again, while the group keeps buying into a subsidiary whose own purchase forecast has not come true for two years running.

Beyond its operating business, Fox Corporation holds a call option on 18.6 percent of FanDuel, the largest U.S. sports-betting operator and a subsidiary of Flutter Entertainment. The strike price was originally fixed at $3.7 billion and has climbed 5 percent a year since — by June 30, 2026 it stood at about $4.7 billion, equal to 40 percent of Fox's entire shareholders' equity. It is a right, not an obligation — Fox does not have to exercise it, and no pro-forma calculation for the Roku acquisition assumes it does.

On top of that, Flutter cannot take FanDuel public without Fox's consent — a lever beyond the call option itself. For context on the exposure: Fox's direct stake in Flutter (roughly 2.5 percent of its shares) already produced a $761 million valuation loss in fiscal 2026 and was the main reason reported GAAP net income fell. A future exercise of the FanDuel option would tie up the balance sheet further — at a time when cash is already getting tighter because of the Roku financing.

As of August 17, 2026, 37,002,060 Class B shares held by the Murdoch holding company LGC Holdco were pledged as collateral for loans — that is 16.8 percent of all Class B shares and 43.3 percent of LGC's entire position. The background: in September 2025, LGC Holdco bought out the departing Murdoch siblings (Prudence MacLeod, Elisabeth and James Murdoch) for roughly $1.99 billion combined, partly financed with a secured loan. Proxy adviser ISS subsequently recommended voting against four Fox directors.

A new stockholders agreement dated September 8, 2025 caps the Murdoch side at no more than 44 percent of Class B votes, so the pledged 16.8 percent sits inside a control block that is already capped. Should the loan turn distressed or require additional collateral, that would surface in a Schedule 13D/A or Form 4 filing before it shows up in the stock price.

In August 2025 a U.S. law firm published a presumably AI-generated false press release that tied a nearly two-year-old filing to a 20 percent drop in the Pro-Dex share price. Pro-Dex turned the incident into its own risk factor in the quarterly report — a risk practically unheard of in SEC filings so far.

On top of that, the crypto token Polkadex trades on crypto exchanges under the very same PDEX symbol. Pro-Dex itself warns that automated alerts or trades could be misdirected — and with only about 1.7 million freely tradable shares, that can move the price.

From selling its Monogram stake to Zimmer Biomet, Pro-Dex holds 2,212,378 non-tradeable contingent value rights. If all five milestones are met — including an FDA 510(k) clearance and revenue targets — up to $12.37 per right pays out: in total up to $27.4 million, about 14 percent of the $198 million market cap (August 31, 2026).

The first milestone already delivered $2.3 million in the third quarter of fiscal 2026. The remaining four are not on the balance sheet, appear in no financial metric — and, as Pro-Dex stresses, come with no guarantee whatsoever.

Pro-Dex’s securities portfolio is run by an investment committee chaired by chairman Nicholas Swenson, who per a beneficial-ownership filing (Schedule 13D/A) of June 26, 2026 controls 31.88 percent of all shares himself. Per the quarterly report, the company explicitly invests “surplus operating capital or borrowed funds” — including in stocks the directors may own privately or through their funds.

Concretely, $986,000 sat in Air T, Inc. as of March 31, 2026 — where the same Swenson serves as CEO and chairman, and another Pro-Dex board member works as chief of staff. 38 percent of pre-tax profit in the first nine months of fiscal 2026 came from securities gains, not from the surgery business.

STIF's 2025 annual report lists two further acquisitions in its "events after the balance sheet date" section, both taking place between December 31, 2025 and the report's publication on April 30, 2026: through its Belgian subsidiary Torino Holding BV, STIF acquired a majority stake in Danish explosion-protection specialist SAFEVENT — described as a continuation of the 2025 Stuvex integration to strengthen its presence in Northern Europe. On April 8, 2026, STIF also raised its stake in BOSS Products UK by 30 percentage points to 100 percent. That means STIF completed or expanded four acquisitions in a little over a year (Stuvex, BOSS Products USA, SAFEVENT, BOSS Products UK) — a pace that fits the gap between organic and reported growth described elsewhere in this analysis: without a steady stream of new deals, the growth story would increasingly have to rest on an organic pace that was just 10 percent in H1 2026.

STIF's 2025 annual report documents two off-market block sales by the Burgos founding family within a few months: on July 31, 2025, Manuel Burgos (founder, Deputy CEO) sold 70,000 shares and his daughter Valérie Burgos sold 25,000 shares — 95,000 shares combined (1.85 percent of capital). On October 22, 2025, a second sale followed: holding company JB Participations, controlled by chairman and CEO José Burgos, sold a further 250,000 shares (4.9 percent of capital). In total, 345,000 shares (6.72 percent of capital) changed hands, without the report disclosing a sale price — at share prices between roughly €50 and €90 during that period, an estimated value of several million euros. After both sales, the Burgos family still holds a majority of voting rights (67.6 percent, on 51 percent of capital) — the sales were partial disposals from a highly concentrated position, not a change of control. They took place months before the July 22, 2026 revenue release made the organic growth slowdown public.

In the first half of 2026, cash cost per pound of copper net of by-product credits stood at minus $0.03 — a year earlier it was plus $0.70. Arithmetically, Southern Copper no longer carries its own copper mining: silver, molybdenum and zinc pay for all of it and leave something over.

The flip side sits in the same reconciliation table: before by-product credits, cost rose from 208.2 to 229.7 US cents per pound. The company's own cost base is getting more expensive. The sign depends entirely on metal prices it does not influence — the average silver price in the half-year was more than twice the prior-year level.

Since the second quarter of 2024 Southern Copper has paid a growing part of its distribution not in cash but in treasury shares. The quarterly report for the period ended June 30, 2026 puts that holding at 50,267,580 shares — at the end of 2025 it was still 65,497,804. In a single half-year, 15,222,224 shares flowed out.

There is practically no resupply: according to the 2025 annual report, no buybacks have been made since the third quarter of 2016, and the remaining authorised volume covered only around 0.6 million shares. At the pace of the first half of 2026 the holding lasts about three more half-years, that is until roughly the start of 2028. The rate declared for August 27, 2026 — 0.0120 shares per share — sits above the half-year average of 0.00925, however, so if it keeps rising the holding runs dry sooner. At that point the board must either pay cash, issue new shares or cut the distribution.

Per the summary of the June 11, 2026 analyst conference, United Integrated Services' geographic revenue split shifted dramatically within one quarter: for full-year 2025, Taiwan still accounted for 66 percent of revenue, the US for 25 percent, and other countries for 9 percent. In the first quarter of 2026, the US share, at 56 percent, had already overtaken Taiwan's 40 percent (4 percent other). Management indicated the US share could reach roughly half of consolidated revenue for the full year 2026, driven by construction progress on a US customer's second expansion phase in Arizona. Operationally that is good news — more orders, higher utilization — but it also means US-dollar instead of Taiwan-dollar revenue at a company that still reports in Taiwan dollars, US wage and construction-cost structures instead of Taiwanese ones, and growing exposure to US permitting, tariffs, and the political durability of US semiconductor incentive programs.

The 2025 annual report of United Integrated Services lists three customers in its "Major customers" section (page 106) that together accounted for 91.82 percent of consolidated revenue in 2025: "Customer A" with TWD 27,073,626 thousand (40.96 percent), "Customer B" with TWD 16,814,377 thousand (25.44 percent) and "Customer C" with TWD 16,799,107 thousand (25.42 percent). All three are listed as "None" under "Relationship with the issue[r]" — none are related parties. The concentration actually worsened versus 2024, when the largest customer stood at 36.74 percent and "Others" at 12.80 percent — in 2025, "Others" fell to just 8.18 percent. The report does not name any of the three customers; independent press coverage and investor-conference summaries identify TSMC and Micron Technology as major customers of United Integrated Services, but that does not let a reader map either name onto "A," "B," or "C."

As of August 5, 2026: In the Q&A section of the second-quarter 2026 earnings call, analysts specifically pressed management on component supply for the second half. CFO Eric Chen and COO Linda Tsai gave a nuanced answer: supply for commodity parts such as memory and SSDs had eased noticeably. Printed circuit boards (PCBs) were the exception — management expects the situation to tighten further in the fourth quarter of 2026 because of upstream material shortages and some suppliers discontinuing production. Certain low-core-count CPUs were also occasionally tight, though management called that situation manageable.

Notably, this risk does not show up as a line item in any quarterly report published so far — neither Q2 2026 revenue nor margin reflects a PCB shortage. It is a pure management statement from the free-form Q&A, not an effect documented in the financial statements. That is exactly what makes it a side find: record order intake ($574 million in July 2026, book-to-bill 1.64) promises a strong second half — but whether Advantech can actually source enough PCBs to ship those orders in the fourth quarter was still an open question at the time of this analysis.

As of March 31, 2026: The shareholder list in the 2025 annual report is topped not by an investment fund but by ASUSTEK COMPUTER Inc. — the Taiwanese PC and component maker holds 113,483,106 shares, or 13.07 percent of Advantech. Right behind it are two investment vehicles the same table itself identifies as linked to Chairman K.C. Liu: K and M Investment Co., Ltd. (100,651,794 shares, 11.59 percent) and AIDC Investment Corp. (99,746,136 shares, 11.49 percent, represented by Mary Chang). K.C. Liu personally holds a further 26,993,951 shares (3.11 percent), and the nonprofit Advantech Foundation holds 24,543,548 shares (2.83 percent) — the annual report lists the Foundation, too, as a related party to K.C. Liu.

Added up, the four Chairman-linked positions total roughly 29 percent — almost double ASUSTEK's stake. For public shareholders that remains structurally relevant: Advantech is broadly held on paper (35,111 shareholders per the annual report, with foreign institutions at 32.59 percent), but control clearly rests with the founding family plus a single outside shareholder from the same industry.

Global Unichip carried almost no debt for years — the balance sheet as of June 30, 2026 shows little more than a small short-term item. On August 19, 2026, the board approved two financings at once, per matching coverage from cnyes.com and BigGo Finance: a syndicated loan led by Mega International Commercial Bank of NT$40 billion (with an option to upsize by up to 25 percent to a maximum of NT$50 billion), a five-year term from first drawdown, and the company's first unsecured overseas convertible bond of up to $800 million (roughly NT$25.5 billion), a planned five-year tenor, denominated in $200,000 units. Combined, that is up to NT$65.5 billion (roughly $2.1 billion) — more than the company's entire NT$37.4 billion balance sheet as of June 30, 2026.

Stated purposes are strengthening medium- to long-term working capital (the loan) and securing foreign-currency funding for materials (the bond) — consistent with the capacity build-out for expected CPU and AI-inference mass production. Weighting the mix toward bank debt over pure equity is meant to avoid a large one-time dilution, according to the coverage — but the convertible bond remains a dilution instrument in its own right: if converted below the prevailing share price, new shares get created. Final terms (coupon, conversion price) had not been published as of this analysis's cutoff.

Asked about hyperscaler cash flow and capital returns, Lanford Liu, Vice President of Corporate Investment, used the July 30, 2026 earnings-call transcript to volunteer a third risk absent from any prior question: American public opinion. Recent surveys, he said, show roughly 55 to 56 percent of Americans express significant concern about AI, and more than 70 percent of respondents are strongly opposed to having an AI data center nearby. More than ten U.S. states are considering legislation that could slow data-center construction, Liu said, and New York's governor has already signed an executive order temporarily halting projects above 50 megawatts. Notably, this assessment came not from an outside critic but from Delta's own investor-relations team — for a company whose data-center business, by its own account, now exceeds half of group revenue.

In the online Q&A portion of the July 30, 2026 earnings call, CFO Beau Yu disclosed a balance-sheet change absent from prior press coverage of the quarter: Delta shortened the depreciation period for newly acquired AI-related manufacturing equipment from a previous three-to-five years to a uniform three years — agreed with the company's auditors and applied since April 2026. Chairman Ping Cheng explicitly framed the move as a prudence measure given the risks of the current AI cycle. A shorter depreciation life means higher annual depreciation charges on the same investment, weighing on reported profit — and it applies to an equipment base that Chairman and CEO Ping Cheng said should grow to roughly TWD 70 billion in 2026 (2025 actual: TWD 46.1 billion).

Per the July 30, 2026 investor presentation (p. 6), revenue at the Automation segment (industrial and building automation) grew 15 percent year over year to TWD 15.5 billion in the second quarter of 2026 — but segment profit swung to a loss of TWD 443 million over the same period, after the segment had been profitable a year earlier. Chairman Ping Cheng, on the same day's earnings-call transcript, blamed a memory-chip shortage that prevented planned shipment volumes and drove up purchasing costs. Delta has since redesigned the affected products with a second memory-chip source; the backlog should be cleared in the second half, and the segment should "perform better than in the first half" and "should not remain loss-making."

Note 18 "complex financial instruments" of the half-year report filed August 14, 2026 discloses a convertible redeemable preferred stock liability with a carrying value of KRW 207,259 million. The sole holder, per the shareholder register, is a party named "Polish Company Limited," which holds all 1,175,647 preferred shares — 10.16 percent of the 11,565,295 total shares outstanding (common and preferred combined). The shares were issued October 8, 2024; the report does not disclose who is economically behind the "Polish Company Limited" name. The conversion price is KRW 170,119 per share — roughly 37 percent of the common-share price of KRW 461,500 (August 28, 2026), meaning the common price sits about 171 percent above the conversion price. A detail table in the same report (Section I.4) lists the conversion window as October 8, 2025 through October 7, 2034; Note 18 instead states the right becomes exercisable on October 8, 2027 — the same date Section I.4 gives for the holder's separate redemption right. Per the more specific Section I.4 table, conversion has therefore already been possible since October 2025.

The report states the resulting share count directly, not as a calculation: 1,175,647 new common shares (Section I.4, a 1:1 conversion ratio) — more than 11 percent above the 10,389,648 common shares outstanding today. Because the conversion price sits well below the market price, conversion rather than a cash redemption is the economically obvious path for the holder. For existing shareholders, that means the earnings-per-share figures behind today's growth story do not yet account for a dilution that, per the more specific report table, may already be due.

The 2026 EPS consensus estimate (TWD 91.79) comes from just one analyst according to fundamental data; seven days earlier, the same estimate stood at TWD 76.60 — an upward revision of roughly 20 percent within a week. The 2027 estimate (TWD 112.66) rests on two analysts. Both figures drive the widely cited forward price-to-earnings ratios of roughly 25 (2026) and 20 (2027), against a trailing P/E of roughly 39 — a base this thin means a single model change by a single analyst can shift the entire valuation debate around the stock without anything actually changing at the company.

In Q1 2026, Syntec's gross margin reached 48.77 percent per fundamental data — a high for the six quarters under review. In Q2 2026 it fell to 43.62 percent, a drop of 5.15 percentage points in a single quarter, even as revenue jumped from TWD 4.86 billion to TWD 8.26 billion in the same period. At the same time, operating margin kept rising, from 25.78 to 28.93 percent — a new high for the series. This divergence (falling gross margin, rising operating margin) points to a pure fixed-cost scale effect rather than a broader profitability problem — but it also confirms that product mix or input costs shifted noticeably as volume roughly doubled. None of the sources reviewed (the May 22, 2026 investor conference, fundamental data) explains the gross-margin decline in detail; the conference itself predates the release of the Q2 figures.

Between Dec. 31, 2025 and June 30, 2026, Syntec's share count rose from 66,495,811 to 71,760,000 according to fundamental data — up roughly 7.9 percent in six months, of which about 5.0 percent came in the first quarter of 2026 alone. The May 22, 2026 investor conference mentions in passing, in an analyst question about return on equity, that the company remained profitable at a 37.7 percent ROE "after a cash capital increase" (進行現金增資後) — without naming a subscription price or share count for that increase. That differs from the well-documented capital increase ahead of the IPO in September 2025 (6,320,000 shares at a provisional TWD 630, proceeds of roughly TWD 4.79 billion), for which subscription price, share count and proceeds are individually documented across several Taiwanese business outlets. This research found no comparably detailed public disclosure for the capital move made in the first half of 2026 — a gap only the official MOPS filing could close.

Several trade outlets — including Taiwan's CommonWealth Magazine and the analysis site SemiWiki — independently report that Chroma ATE is the sole supplier of the system-level test (SLT) equipment that Nvidia's packaged AI chips must pass through before shipping. The company itself does not name customers in its own reports, citing confidentiality — its July 30, 2026 earnings presentation speaks only generally of "SLT from AI / HPC / ASIC" as a growth driver. Such a concentration on a single semiconductor customer, if accurate, would be a material concentration risk: any delay or shift in Nvidia's own product cadence would hit Chroma disproportionately — even though the company says it now also tests AMD and Google chips.

The product-mix tables in the earnings presentations show a collapse that no pure AI growth story mentions: revenue from Turnkey Solutions (electric-vehicle and battery test and automation systems) fell from NT$685 million in full-year 2024 through NT$444 million in 2025 to just NT$106 million in the first half of 2026 (down 61 percent year over year) and NT$53 million in the second quarter of 2026 alone (down 57 percent). Measured against parent-company revenue, the segment's share fell from 4 percent in 2024 to just about 1 percent in the first half of 2026. Effectively all of Chroma ATE's reported growth now hinges on two AI-adjacent segments (Test Instruments & ATS, and semiconductor/photonics test) — the broader end-market diversification the company had two years ago is gone.

A single sentence tucked beneath the results table in the Oct. 30, 2025 earnings presentation reads: "Gain from disposal of residential apartment held for sale (to employees) of NTD 3,185 million." Of the NT$5.066 billion net income attributable to shareholders in the third quarter of 2025, NT$3.185 billion — nearly 63 percent — came from the sale of a single staff apartment, not from the core business. Operating income that same quarter was only NT$1.848 billion. Anyone crediting the widely quoted 253 percent year-over-year profit jump entirely to the AI test business is falling for a one-time item.

In its Dec. 4, 2024 investor conference presentation, Jentech said the expansion of Dayuan Plant 1 (10,579 additional square meters of floor space) was "expected to be completed by Q4 2025." In the follow-up presentation on Nov. 28, 2025, the same expansion (by then enlarged to 15,653 square meters) read: "expected to be completed by Q1 2026" — a one-quarter slip. At the same time, the completion date for phase 1 of the larger Aerotropolis plant was pulled forward from "Q4 2027" (December 2024) to "Q1 2027" (November 2025) — a mixed picture of delay on the smaller project and acceleration on the bigger one. Together, the two projects are meant to more than double Taiwan floor space from 46,413 to roughly 99,554 square meters — an expansion the entire growth story depends on.

The investor conference presentation from Nov. 28, 2025 breaks revenue down into five product lines: while Thermal Solutions grew 77 percent year over year in Q3 2025, communication connectors shrank 35 percent and leadframes 9 percent in the same period — both product lines that carried a far larger share of revenue before the AI story (leadframes alone made up 19.1 percent of revenue in Q1 2024, versus just 9.4 percent most recently). The company itself assigns these two segments the lowest confidence rating anywhere in its own 2026 growth outlook (3 and 4 out of 5 "smileys", versus 4 to 5 out of 5 for heat spreaders and liquid cooling) — a detail that never shows up in a pure AI-cooling narrative but comes directly from the company's own presentation.

According to fundamental data (as of Aug. 28–30, 2026), Jentech's annualized 30-day volatility stands at 86.4 percent and its 250-day volatility at 72.9 percent — extreme readings for a cooling-hardware supplier. The same data source shows a beta of just 0.063 — near zero, meaning the stock barely correlates with the broader market. That combination is the real FOMO driver: the price action is almost entirely company-specific, not something that can be read off a market index. A look at ownership offers a plausible explanation: only about 59 percent of the 147.2 million shares are considered free float (86.3 million shares), against an insider ownership rate of roughly 38 percent. That concentrated a shareholder base means relatively modest trading volume can move the price sharply — on Aug. 17, 2026, the stock traded, per price history, with essentially no intraday range at all (open, high, low, and close identical).

On February 19, 2024, external attackers obtained employee credentials through social engineering and used them to deploy ransomware on internal servers — an incident Sunonwealth discloses itself in the 2024 annual report and classifies as not materially significant, despite holding ISO 27001 certification. A few months later, on June 14, 2024, the shareholders' meeting re-elected the board of directors — with the result that Chairman Ching-Shen Hong now also serves as President (general manager). The annual report itself frames this as continuity without a control risk; from an outside governance perspective, it means that oversight and operating management sit with the same person, with no disclosed external selection process for the combined role.

Between March 31, 2026 and June 30, 2026, Sunonwealth's shares outstanding grew from about 275.4 million to about 285.4 million according to fundamental data — a jump of 3.7 percent in a single quarter, well above the average annual increase of prior years (which, per the 2024 annual report, mostly came from the conversion of corporate bonds). We could not find a publicly available explanation for this specific jump — a bond conversion, a capital increase or a stock dividend — in the sources available; the 2024 annual report had explicitly stated, for the 2024 profit distribution, that no stock dividend was planned. That may have changed for fiscal year 2025 — the next annual report should show it.

The May 21, 2026 investor presentation shows cold plates, coolant distribution units (CDUs) and full rack systems for AI data centers — yet by the company's own timeline, these products are set to deliver a meaningful revenue contribution only in the second half of 2026; the first quarter of 2026 saw only initial deliveries. Fitting that picture, revenue growth has already slowed: from 20.6 percent year over year in the first quarter of 2026 to roughly 16.2 percent in the second quarter (our own calculation from fundamental data: TWD 5,418.6 million in revenue versus TWD 4,663.3 million in the prior-year quarter). Net income, by contrast, grew 84.5 percent in the same quarter — revenue and profit moved in different directions, carried by a rising margin rather than higher unit volume.

Taiwan Union Technology is not led by an externally recruited executive after its most recent chairman transition — leadership stayed inside the founding Hsin family: press reports say outgoing chairman Hsin Chung-tao (辛忠道) stepped down and his deputy chairman Hsin Chung-heng (辛忠衡) moved up, with no outside candidate holding the role in between. Taiwanese media describe the Hsin family as one of the most influential families in the city of Tainan, with ties to politics and business spanning generations. A family-internal leadership change is not by itself a warning sign — continuity can be an advantage for a company mid-expansion. But it does mean control over a company worth roughly TWD 430 billion in market value stayed within a single family, without a documented outside selection process.

For Taiwan Union Technology's seemingly high price-earnings ratio to compute away, a lot has to go right: the analyst consensus (as of August 16, 2026; 12 analysts for 2026, 13 for 2027) expects full-year 2026 earnings per share of TWD 36.69 — since the first half alone already delivered TWD 12.4, the second half would essentially have to double the pace of the first. For 2027, the consensus estimate stands at TWD 75.10 — a further gain of roughly 105 percent over the 2026 estimate. Only once both jumps materialize does the trailing price-earnings ratio of roughly 75 (on the last four reported quarters) fall to a range around 20 that would look reasonable for a materials maker. If either jump falls short, the stock stays expensive by conventional metrics.

In the second quarter of 2026, Taiwan Union Technology earned TWD 2.34 billion after tax — the best quarter in the company's history. Yet operating cash flow ran negative, not positive: an independent write-up of the July 30, 2026 earnings call puts second-quarter operating cash flow at minus TWD 716 million, with a further TWD 940 million in capital expenditure on top. The reason lies in growing inventory and receivables plus ramp-up costs at the new Thailand plant, whose second construction phase ran roughly two months late. A profit that has not (yet) turned into cash is not by itself a red flag for a company mid-expansion — but it is a finding the income statement alone does not show, and if the pattern persists it makes the already-announced convertible bond (TWD 10 billion) more important than a pure profit view would suggest.

In its third-quarter 2025 report, Barrick writes verbatim: "After nearly twelve months fatality free, unfortunately three of our colleagues lost their lives in recent months." A previously recorded lost-time injury at the Kibali mine was reclassified as a fatality after the employee later succumbed to injuries; on September 29, 2025, an employee died at the Goldrush underground mine in Nevada, and on October 21, 2025, another died at the Bulyanhulu mine in Tanzania — three different countries, three different operations, eight weeks.

In its second-quarter 2026 report, the company reiterates that it invested $90 million in safety technology in 2026 and is focused on "visible leadership, consistent engagement, and a stronger focus on critical risk management." That is an explicit, though unquantified, commitment to improvement — whether it is working will only show up in the accident record of coming quarters, which Barrick does not currently publish as a consistent, quarter-by-quarter metric.

The report for the second quarter of 2026 (August 10, 2026) contains a sentence easy to skim past: "Mark Hill will be the CEO of the new company" upon separation. That refers to the planned spin-off of Barrick's North American gold assets, targeted for completion by the end of 2026. Hill had only been confirmed as permanent President and CEO of Barrick in February 2026 — less than five months earlier — after steering the company through Mark Bristow's abrupt September 2025 departure and helping negotiate the resolution of the Mali crisis.

For the remaining, internationally focused Barrick Mining Corporation, that means: after the first leadership change within a year (Bristow → Hill), a second change looms at the latest once the North American IPO closes — this time for the company that keeps the name and the international mines, including operations in Mali, Tanzania, the Dominican Republic, and the planned Reko Diq copper project in Pakistan. Who will lead that new team had not been named as of this analysis's data cutoff.

On September 29, 2025, Barrick Mining Corporation disclosed via a mandatory filing to the SEC that Mark Bristow was stepping down as President and CEO "after nearly seven years," with COO Mark Hill taking over as interim CEO effective immediately. The press release gives no reason — not health, not personal, not business-related. Chairman John Thornton is quoted thanking Bristow for his "leadership" and wishing him "the very best for his future," nothing more. The departure came in the middle of a stretch the company itself described as delivering "solid Q2 operating performance, strong cash flows, an enhanced quarterly dividend and strong share price performance" — explicitly not a period of weakness.

What successor Mark Hill himself admitted six weeks later is notable: "Since assuming interim CEO responsibilities at the end of September, I have met with our teams across the globe to review performance and assess what we can do differently at Barrick." A new chief executive who explicitly announces he is reviewing "what we can do differently" points to more than a smooth handover — even though none of the press releases ever say so directly.

Anyone who googles Kongsberg's "earnings per share" headline in August 2026 sees a decline: NOK 3.73 in the first half of 2026 versus NOK 4.43 a year earlier — a 16 percent drop that looks like a weakening business. The opposite is true. According to the group income statement, earnings per share from the continuing business rose from NOK 1.66 to NOK 2.91 over the same period — a 75 percent increase. The difference: last year's figure still carried heavy profit contributions from the now-divested Kongsberg Maritime business (NOK 2.77 per share from "discontinued business" alone in the first half of 2025, versus just NOK 0.83 in the first half of 2026, since Maritime now only counts through April 23).

For any comparison against prior-year figures or analyst estimates, the takeaway is the same: whoever carries forward the total EPS line unreflectively sees a decline that isn't one. The relevant number to track is the separately disclosed "earnings per share from continuing operations" line — and that one shows a clear gain.

Kongsberg is celebrating a missile boom in 2026: Canada, Germany and the U.S. Air Force together ordered NOK 10.9 billion worth of Joint Strike Missiles (JSM) in the second quarter alone, and the order backlog of the Missiles & Aerostructures segment grew to NOK 68.1 billion. All the more surprising, then, is a single, unassumingly placed line in the quarterly report's notes: the Norwegian government itself withdrew Kongsberg's export license for a NSM missile contract with Malaysia announced in 2018 — contract value €124 million, or roughly NOK 1.4 billion. The group confirms this verbatim and adds that "related accounting assessments have already been reflected in the financial statements as of the second quarter" — so there is already a real balance-sheet impact, though the exact size is not disclosed.

For investors, the individual Malaysia case matters less than the precedent: if the group's own home state — which shares in more than 50 percent of Kongsberg's stock — can retroactively pull back a once-granted export license, the same risk cannot be ruled out for the far larger current contracts (Canada NOK 4.7 billion, Germany NOK 3.5 billion, U.S. Air Force NOK 2.7 billion, all for the same JSM missile). At this stage the group itself offers no further detail — "due to ongoing negotiations between the parties."

Update: Press reports now say Malaysia's government has put a number on its claim — around €226.1 million (roughly 1.06 billion ringgit), consisting of €129.86 million for payments already made and €96.26 million in follow-on costs — more than 1.8 times the original contract value. Initial talks were scheduled for mid-August 2026; as of August 27, 2026 no settlement had been reported. Kongsberg itself has not confirmed this figure.

Note 13 to the fiscal 2026 annual report sets out two proceedings side by side that together touch more than 110 million dollars. First, Ubiquiti has been disputing fiscal 2015 and fiscal 2016 with the US Internal Revenue Service for years. After a Notice of Deficiency dated 3 August 2022 the company petitioned the US Tax Court; on 22 January 2024 the assigned judge rejected both sides' motions for summary judgment, and Ubiquiti has been awaiting a trial date since. Its own estimate of the incremental tax liability: approximately 50.0 million dollars, excluding potential interest and penalties. Second, the Hong Kong Inland Revenue Department is auditing the years 2010 to 2020; between fiscal 2018 and fiscal 2025 Ubiquiti paid 60.9 million dollars in deposits there, of which 60.3 million (net of currency effects) sits in other long-term assets as refundable.

The accounting treatment is the striking part. On the IRS matter the filing says verbatim: "As the Company believes that the tax originally paid in fiscal 2015 and fiscal 2016 is correct, it has not provided a reserve for this tax uncertainty." The same logic applies to the Hong Kong matter. That is permissible, but it pushes the entire exposure into the future: an adverse outcome would hit the income statement in full and without preparation. Measured against cash and short-term investments of 611.2 million dollars at 30 June 2026, this is about 18% of the cash pile.

On 21 August 2025 Ubiquiti's board approved a share repurchase programme of 500 million dollars, originally set to expire on 30 September 2026. The notes to the fiscal 2026 annual report contain the sentence that puts the programme in perspective: "During the year ended June 30, 2026, the Company did not make any repurchases under the 2025 August Program." Not a single share was bought back under the programme during the entire fiscal year. On 20 August 2026 the board nevertheless extended it to 30 September 2027.

Scale is the point here. 500 million dollars equals roughly 52% of fiscal 2026 net income (960.3 million) and about 82% of total cash and short-term investments at 30 June 2026 (611.2 million). More importantly, it would meet a very thin market: because founder Robert J. Pera holds roughly 93% of the stock, only about 4.2 million shares trade freely — worth roughly 2.5 billion dollars at the 28 August 2026 close. A fully executed programme would therefore amount to about a fifth of the tradable float. Whether the authorisation turns into purchases will show up in the repurchase table of the next quarterly report (10-Q). So far it shows zero.

On January 19, 2024, Chenbro issued an unsecured convertible bond with a face value of NT$1 billion (actual proceeds: NT$1,096,916,000, issued at 109.69 percent of face value) — at 0 percent annual interest, maturing three years later on January 19, 2027. The conversion price is NT$266.60 per share (originally NT$271.00 at issuance). As of the annual report's print date (April 8, 2025), not a single bond had been converted. At a share price of NT$990 (August 28, 2026) — more than three and a half times the conversion price — the bond is now deep in the money: fully converted, it would create up to 3,750,938 new shares (NT$1 billion divided by NT$266.60), about 3 percent of the 125,619,146 shares currently outstanding. Measured against total short- and long-term financial debt of NT$2.94 billion (December 31, 2025), the bond makes up roughly 37 percent of interest-bearing liabilities.

The 2024 annual report lists shareholders holding more than 10 percent of the company: alongside Chairwoman Mei-chi Chen herself (8.00 percent directly, as of March 31, 2025), it names Peng Wei Investment with 12,180,000 shares (10.07 percent). A footnote to the shareholder table clarifies who actually owns those shares: "Chairperson Mei-chi Chen held the shares by nominee arrangement." Together, the two disclosed positions already put more than 18 percent of the shares under her control — a link a reader would not spot from the plain ownership list alone. The same report also discloses that director Tsun-yen Lee (4.39 percent directly) is married to Feng-ming Chen (13,191,433 shares, 10.90 percent) and is listed as a "relative-in-law" of the chairwoman. Four named positions in the 2024 annual report add up to roughly 33 percent of the shares, concentrated in a single family circle.

Amphenol acquired the connectivity and cable business of Vistance Networks, formerly CommScope Holding Company, on January 9, 2026. Of the $10,735.8 million in net assets acquired, $6,958.0 million was allocated to goodwill and $3,289.0 million to other intangible assets — together roughly 95 percent. The Form 10-Q for the period ended June 30, 2026 flags this explicitly: the allocation rests on preliminary assessments, the acquisition accounting is "subject to final adjustment," and the final values may differ.

For investors this is not a formality but a profit question. Goodwill is not amortized on a schedule; intangible assets are. The acquired intangibles carry a weighted average useful life of roughly 16 years — proprietary technology 14 years, customer relationships 16 years, trade names 23 years. If the final allocation shifts value out of goodwill and into intangibles, future scheduled amortization rises and reported profit falls, with nothing whatsoever changing in the business. One more detail: the entire goodwill amount was assigned to the Communications Solutions segment, and per the filing none of it is expected to be deductible for tax purposes.

The China tax matter cost Amphenol a total of $390.0 million in the first half of 2026: $230.0 million of tax payment notices, paid in full during the second quarter, plus $160.0 million of additional tax obligations from a reassessment of prior years. That much is in the earnings release. Far less noticed is a sentence two paragraphs further into the Form 10-Q for the period ended June 30, 2026: as of that date, the balance of unrecognized tax benefits, including penalties and interest, stood at roughly $386.1 million — the amount that would hit the effective tax rate if it were recognized.

The company also quantifies how much of that could be settled soon: over the next twelve months, audit activity could be completed and statutes of limitations could close on unrecognized tax benefits of approximately $49.3 million. Amphenol states it is subject to income tax examinations for tax years 2017 and after and has "numerous audits underway at various stages of completion." This matters for earnings power: the adjusted effective tax rate for 2026 has already been raised to 27.0 percent from 24.5 percent in the prior-year half, and the company attributes that both to this matter and to a continued shift of income into higher-tax jurisdictions.

When Quanta announced Taiwan's largest share sale in 19 years on July 29, 2026 — 49 million Global Depositary Shares at $44.17, $2.16 billion, 245 million new ordinary shares against roughly 3,854.5 million shares then outstanding, about 5.97 percent dilution — the obvious press narrative was AI server capacity expansion, given the billions in new California and Mexico facilities announced weeks earlier. Quanta's own stated purpose was more mundane: the proceeds would fund purchases of materials in foreign currencies — working capital, not new factories (Bloomberg, "Nvidia Partner Quanta Raises $2.2 Billion From Share Sale," July 29, 2026; Yahoo Finance/Reuters, July 29-30, 2026).

The distinction matters: a company raising fresh equity to fund growth investment tells a different story than one that can no longer cover routine raw-material purchases out of operating cash flow — especially in the same reporting period gross margin had fallen to a multi-year low (4.78 percent in the first quarter of 2026). The offering also priced at a 7.3 percent discount to the prior close, and the stock lost 9.9 percent on announcement day — a sign the market did not wave the dilution through.

On August 6, 2026, according to the subsequent-events note in the Q2 2026 consolidated financial report, Mega Union's board approved two unsecured convertible bonds — one for up to NT$1,300,000 thousand, a second for up to NT$1,200,000 thousand, together NT$2.5 billion, both at a 0 percent coupon and a five-year term. The proceeds are earmarked for reinvestment in overseas subsidiaries, purchasing and renovating plant buildings, building warehouses, and constructing offices and equipment — above all for the new Arizona site, for which the board approved a cost ceiling of up to NT$1.95 billion on March 12, 2026; the land alone, at NT$375,241 thousand, was already fully paid as of June 30, 2026.

What stands out: as of June 30, 2026, the group held NT$7,448,057 thousand in cash against only NT$728,195 thousand in combined short- and long-term bank debt — a net cash position of roughly NT$6.7 billion. Mega Union could, on paper, have funded the planned investments from its own cash. Instead the board chose a convertible instrument that creates new shares upon conversion — one more step in a series that has already grown the share count from 25.3 million (2017) to 99.7 million (2026). The final terms (conversion price, issuance date) are not yet set, pending regulatory approval.

Mega Union's 2025 annual report discloses the concentration itself, in a table worth reading twice: the largest customer, anonymized as "Company T," bought NT$7,877,768 thousand worth of goods and services from Mega Union in 2024 (76.48 percent of revenue), NT$13,064,633 thousand in 2025 (77.25 percent) — and, in the first quarter of 2026 alone, NT$3,894,243 thousand out of total revenue of NT$4,706,862 thousand, or 82.74 percent. The ten largest customers combined reached 95.56 percent of group revenue in 2025 (2024: 94.90 percent) — practically the entire business rests on a handful of customers, the lion's share on a single one.

Mega Union itself explains why in its risk section: the customer is "the world's leading semiconductor company," with a capital budget of roughly US$30 billion in 2024 and an estimated US$38 to 42 billion in 2025 — and the company has supplied its ultrapure-water and wastewater systems since 2008 at the Hsinchu Science Park, later expanding to Central and Southern Taiwan, completing more than 50 new-fab projects. The filing never names the customer — but Taiwanese financial media and investor forums independently and consistently identify "Company T" as TSMC, which matches the investment scale cited.

Founder Lin Tsung-chi was born in 1941 into a poor family in Kaohsiung, according to press reports (bnext.com.tw, retrieved 2026-08-29) started working as a furniture-factory apprentice at age ten, and attended only elementary school. On August 11, 2026 — five days after the record analyst conference and the stock's move past TWD 10,000 — press estimates put him at roughly $16.7 billion and named him Taiwan's richest person. At King Slide's market capitalization of roughly $43.1 billion (August 28, 2026) and an insider ownership share of 40.6 percent per fundamental data, the order of magnitude of that estimate matches what a founder stake of that size would actually be worth at this price.

The investor presentation dated August 7, 2026 shows a net foreign-exchange loss of TWD 2,243,341,000 for the second quarter of 2025 — in a quarter in which revenue (TWD 4,228,734 thousand) and gross profit (TWD 3,277,081 thousand) actually rose slightly versus the prior quarter. Earnings per share still collapsed from TWD 26.35 (Q1 2025) to just TWD 6.45, because the currency loss nearly wiped out an operating profit of TWD 2,966,083,000. Per management's own comments at the August 7, 2026 analyst conference (as summarized by blog.fugle.tw), a one-percent currency swing moves gross margin by roughly 0.5 to 0.6 percentage points — at a company that buys its main raw material (cold-rolled steel) predominantly in Taiwan dollars but increasingly invoices in U.S. dollars.

On December 16, 2025, AES-KY issued a clarification via Taiwan's market observation system after media reports floated a possible profit target of "at least TWD 48 per share" for fiscal year 2027. The company explicitly distanced itself: the cited figures were "analysts' own forecasts," and actual results should be judged "according to the company's official announcements." Eight months later, the picture looks different: the fundamental-data analyst consensus (3 analysts, as of August 29, 2026) already puts fiscal year 2026 EPS at TWD 48.35 — almost exactly the number the company had publicly disowned a year earlier, just arriving a year sooner than originally floated. Based on the first-half-2026 profit (TWD 23.45 per share), a similarly strong second half would get the full year there.

The 2024 annual report discloses the largest suppliers for the last two fiscal years plus the first quarter of 2025. "Company A" (per the report, not a related party) accounted for 66.8% of total purchases in 2023 (TWD 4,748,231 thousand), only 46.6% in 2024 (TWD 2,321,179 thousand) — and just 0.4% in the first quarter of 2025 (TWD 8,632 thousand). The report offers only one sentence: the decrease was "mainly due to the reduction in production demand." Whether AES-KY found a new primary supplier, on what terms, and whether it affected gross margin, remains unaddressed in the document reviewed.

Anyone looking only at the current record quarters misses the recent backstory: in fiscal 2024, Innodisk's revenue rose 7.24% to NT$8,915.6 million — operating profit in the same year fell 14.6% to NT$1,178.5 million, as ongoing investment spending (including the Yilan plant expansion) pressured margins. Net income after tax dropped 3.46%.

The turn only begins in the second quarter of 2025: gross margin climbs quarter by quarter from 25.9% through 32.8%, 33.4%, and 59.1% to 65.0% in the second quarter of 2026 — carried by the global DRAM/flash pricing cycle. This stock's story is therefore not a straight line up, but a cycle: a weak year followed by a historic boom. Whether and how sharply the cycle reverses once global memory supply normalizes remains an open question at this point.

Innodisk's 2024 Annual Report discloses what its business really rests on: Samsung Electronics (South Korea) and Kioxia (Japan) are the primary suppliers for its main raw material, memory chips. Procurement does not run directly but through Taiwanese agents — in fiscal 2024, 37.69% of purchase volume ran through one agent and another 22.23% through a second. Combined, that is 59.92% of total purchases through just two counterparties.

The report itself rates the risk as low ("stable with no risk of supply interruption"). That was true for a buyer's market in 2024. Since 2025/2026 the picture has flipped: the global DRAM and NAND flash market is in a demand boom driven by AI data centers. In a seller's market like this, negotiating leverage with exactly these two agents determines whether Innodisk keeps getting supplied at competitive terms and volumes — a disadvantage against larger buyers with their own framework agreements directly with the memory makers.

In the Q2 2026 earnings release, AVC indicated that second-half 2026 revenue should exceed the first half's NT$98,159 million, because "all product lines are entering mass production." That is a concrete, self-set and checkable bar: the second quarter of 2026 already stagnated versus the first (up just 0.17 percent), so essentially all of the second half's growth still has to be delivered. At the same time, management guided to roughly NT$15 billion of 2026 capital expenditure — with 2027 expected to be higher — and said capacity expansion and investments should be funded from operating cash flow and existing credit lines, without external financing.

For a stock already trading at roughly 45 times earnings, that is not a footnote: a missed second half would call into question not just the growth story but the valuation itself.

In the first half of 2025, server and networking applications made up 48.4 percent of AVC revenue — a year later, in the first half of 2026, that figure is 66.14 percent (NT$64,919 million, up 153.38 percent year over year). Within twelve months, the company's product mix narrowed from a broader spread (notebook/PC OEMs such as HP and Lenovo, automotive, telecom) to a single, highly cyclical investment cycle: the build-out plans of cloud and AI infrastructure operators around NVIDIA platforms such as GB200/GB300, for which AVC was certified in April 2026 according to Taiwanese business media.

This is not a governance or balance-sheet problem, it is a concentration risk: should the pace of cloud capex spending slow — for which there is currently no evidence — that would now hit two-thirds instead of less than half of AVC's business. The Q2 2026 earnings release itself names this shift openly as a growth driver, without offering a counter-scenario for a downturn.

A streaming agreement is generally treated as close to unbreakable — Wheaton buys the right to a fixed share of production for the life of a mine. The 2025 annual report (Form 40-F) shows an exception: in the third quarter of 2025, Chinese mining group CMOC, as part of its acquisition of Lumina Gold, exercised a contractual buy-back option for a third of the Cangrejos gold stream in Ecuador — paying Wheaton $102 million. The deal was financially good for Wheaton: it generated an $86 million gain, since the original capital outlay for that portion of the stream was far lower. Wheaton's claim on Cangrejos gold permanently drops from 6.6% to 4.4% of mine production once 469,000 ounces have been delivered.

For assessing the business model, this is more than a footnote: Wheaton's long-term production guidance (1.2 million gold-equivalent ounces by 2030) implicitly assumes existing streams run as agreed. Change-of-control clauses, like the one triggered in the Cangrejos deal, can undercut that — a mine operator that gets acquired can, under certain conditions, partially buy its way out of a stream. That is not a red flag, but it is a caveat that the headline portfolio size (57 interests) does not show.

For fiscal year 2025, the proxy statement (DEF 14A) disclosed total compensation of $5,350,835 for Chairman and CEO Henry Nahmad. Measured against the company's net income for the same fiscal year ($7.5 million), that equals 71 percent of total profit. In fiscal year 2024, compensation of $5,914,510 actually exceeded that year's net income of $5.65 million — 105 percent. The impact is softened somewhat by the fact that $4.0 million of that pay consists of restricted shares that don't fully vest until November 5, 2040.

The board has also decided that the shareholder vote on executive pay (say-on-pay) will now occur only once every three years instead of annually, a change approved at the December 15, 2025 annual meeting. The finding is material because a single executive's pay routinely claims a double-digit percentage of total company profit — and in one year, more than the entire profit itself.

On July 17, 2026, EVI Industries signed four separate purchase agreements to acquire Sudsies, Inc., a Florida garment-care company — the company's first step from B2B distribution into direct consumer service. Total consideration runs to roughly $34.6 million, about 18 percent of EVI's own market capitalization (roughly $189 million as of August 27, 2026). Unusually, $7,124,778 of that total is allocated solely to purchasing co-founder Jason Loeb's "personal goodwill" — a separate legal structure covering his personal reputation and customer relationships, not the operating business itself.

Closing was contractually expected 30 to 45 days after signing, with a termination right if closing had not occurred by September 1, 2026. The fiscal year 2026 annual report (expected mid-September 2026) will provide the first segment and revenue disclosures for the new consumer business — until then, there's no way to judge whether the new line carries a better margin than the existing B2B distribution core.

On the third-quarter 2025 earnings call the chief executive put the expansion of the Waukesha, Wisconsin plant at "approximately $45,000,000 in capital expenditures" and pointed to completion of the project in 2026. On the full-year 2025 call in March 2026 it was still the "$45 million Waukesha expansion", with phase 2 "on target for completion by the end of 2026". The quarterly report for the period ended June 30, 2026 then carries a different figure: "The Company currently estimates investing approximately $50,500" — 50.5 million dollars, a good 12 percent more. And the completion date reads: "during the first fiscal quarter of 2027".

The overrun appears in the quarterly report only. The new date shows up in the earnings release of August 13, 2026 merely in passing, inside a quote from the chief executive — "the planned completion of our transformative Waukesha expansion in early 2027" — with no indication that it differs from the earlier guidance. There was no separate announcement of the change. It matters: 5.5 million dollars of extra cost equals a third of the entire 2025 operating income, and every quarter of delay is a quarter in which the promised doubling of production capacity cannot yet relieve the compressed margin.

In the first half of 2026, according to the cash flow statement in the quarterly report for the period ended June 30, 2026, Lifeway Foods spent 4.937 million dollars on its own shares — repurchases the filing describes as made "in connection with the sale of Danone USA Public Benefit Corporation's holdings", that is, alongside the placement of the Danone block. In the same six months the business generated only 4.039 million dollars of operating cash flow, while 19.505 million went into property and equipment. The bank closed the gap: 23.0 million dollars were drawn under the revolving credit facility and 1.0 million repaid.

That turns a buyback which looks like strength on the balance sheet into a debt-funded one. The hard number sits in the "Debt Obligations" section: of the 25.0 million dollar facility, 22.0 million was drawn at June 30, 2026 and only 3.0 million remained available — against cash of 7.1 million and a plant expansion that still needs roughly 11.6 million dollars to finish. An equipment financing facility of up to 22.0 million dollars signed in June 2026 stands ready, but was undrawn at the reporting date and depends on the lender accepting collateral documentation.

On July 10, 2026, Niagen Bioscience filed a prospectus supplement (Form 424B5) setting up an "at-the-market" program: through two banks, Canaccord Genuity and Roth Capital Partners, the company can sell new shares worth up to $50 million at any time — against a roughly $255 million market cap, that would be nearly 20 percent of additional share capital if fully drawn. In the very same half-year, the stock-repurchase program launched in November 2025 and expanded to $20 million in March 2026 kept running: $5.1 million of buybacks in the first half of 2026, $2.8 million of it in the second quarter alone.

The quarterly report for the period ended June 30, 2026 states explicitly that, "as of the date of this Quarterly Report on Form 10-Q," the company had "not sold any shares under the ATM Facility" — so for now the dilution risk is purely theoretical. That is exactly what makes it worth watching: if the company draws on the ATM facility in a future quarter while operating cash flow stays weak, that would signal its own cash is no longer enough to fund the new bets (Niagen Plus, NB4168). If the facility stays unused and buybacks continue, that points to a precautionary measure rather than an urgent capital need.

From April 22 to June 21, 2025, Hon Precision ran a buyback program: the target was 3,500,000 own shares at prices between TWD 515.00 and TWD 1,270.00, earmarked for transfer to employees. According to the 2026 meeting handbook, the company actually bought back only 27,000 shares for TWD 20,932,856 — 0.77 percent of the planned amount. The company's stated reason: it bought "in tranches depending on price movement and trading volume, in order to balance market mechanics with shareholder interests," so the buyback was not fully executed. At the time of this program the stock was still trading on the pre-IPO Emerging Stock Market; since the official listing on November 27, 2025 it has multiplied. At TWD 6,490 (August 27, 2026) the price stands at roughly five times the upper end of the buyback range.

The meeting handbook of the 2026 annual shareholders' meeting shows an employee bonus pool of TWD 850,000,000 for fiscal year 2025 (the statutory ratio of 5 percent is calculated on the pre-bonus statutory base under Article 30 of the company's articles and Article 235-1 of Taiwan's Company Act, not on reported net profit), of which at least 40 percent is earmarked for rank-and-file employees. The board also approved director compensation of TWD 80,000,000, paid in cash. For a company that has only been publicly listed since November 27, 2025, that is an unusually generous start: measured against the reported net profit of TWD 12,361,796 thousand, the pool works out to about 6.9 percent — the two percentages differ because they use different bases, not because they contradict each other. Either way, a substantial bonus goes to staff and management before a full calendar year as a public company has even passed.

Between year-end 2025 and July 31, 2026, M-tron's shares outstanding rose from about 2.98 million to 4,346,476 — up roughly 46 percent. Two capital actions drove it, one after the other: first a warrant exercise between late 2025 and January 7, 2026 (582,233 shares, $27.7 million in gross proceeds), which took share count to 3,565,118 (as of the 10-K cover date, 03/16/2026); then a rights offering at $59.00 per share, completed April 27, 2026 (713,362 shares, $42.1 million in gross proceeds), plus roughly 68,000 more shares from option exercises through the end of July. Together, cash grew from $20.9 million at year-end 2025 to $96.2 million as of June 30, 2026 — with zero bank debt.

On the second-quarter 2026 earnings call (08/13/2026), CEO Cameron Pforr confirmed the company had seen "an increase in deal flow" from investment banks since completing the capital raise and "still hope[s] to get a deal done this year." As of this writing, no acquisition has been announced — the fresh capital still sits unchanged in the bank account, diluting every share outstanding before the warrant exercise by roughly 46 percent without yet working productively in the business.

As of March 31, 2026 there were 12,636,605 shares outstanding. The same quarterly report lists 2,298,287 options outstanding at a weighted average exercise price of $4.37, of which 1,647,418 were already exercisable at an average of $2.66. That is 18.2 percent of the shares outstanding — and with the stock trading well above both averages, almost all of it in the money. It already shows in the income statement: the diluted share count for the third quarter was 14,064,470 against 12,617,328 basic.

This is where part of the growth disappears. Over the nine months to March 31, 2026 net income rose 15.1 percent while diluted earnings per share rose 2.0 percent. One more item from the same report: chief executive LeeAnn Rohmann adopted a Rule 10b5-1 trading plan on March 6, 2026 providing for the sale of up to 60,000 shares through June 9, 2027. Such plans are a common, pre-arranged way to sell and say nothing about prospects on their own — but they belong in the same calculation as the options.

On April 1, 2026 — one day after the balance sheet date of the most recent quarterly report — Legacy Education leased roughly 53,000 square feet in Wildomar, California. The Form 10-Q for March 31, 2026 describes the contract as a subsequent event: base term of up to twelve years, occupancy in phases through January 2028, escalating rent starting at about $2.40 per square foot per month plus a pro-rata share of operating costs, property taxes and insurance. At full occupancy that works out, by our own calculation, to roughly $1.5 million of rent a year — about a fifth of fiscal 2025 net income of $7.53 million.

The decisive part sits in a subordinate clause: the report states that the lease has not yet been recognized under the leasing standard and will be recorded in future periods. So the operating lease liability of $14.73 million as of March 31, 2026 is not the full picture. Rent is a fixed cost that does not shrink when fewer students enroll — and in the quarter to March 31, 2026, 1,078 new students started against 1,227 a year earlier. The next quarterly report will show how far the liability jumps.

Without eligibility for federal student aid (Title IV), a U.S. for-profit career college has almost no revenue. That eligibility is not a permanent state but a dated contract with the Department of Education, the Program Participation Agreement. The annual report on Form 10-K for fiscal 2025 names the expiry date for the two largest schools explicitly: “The current expiration date of the program participation agreements for HDMC and CCC is September 30, 2026.” As of March 31, 2026, High Desert Medical College and Central Coast College together accounted for 2,844 of the 3,550 students — roughly 80 percent.

The other two schools hold no current agreement at all, only a temporary one renewed month by month while the regulator reviews the change-of-ownership applications: “Integrity and CCMCC are currently participating in the Title IV Programs under a temporary provisional program participation agreement.” The same report concedes in its risk factors that a reduction in force at the department eliminated “the school participation division that previously oversaw the operations of our institutions” — precisely the unit responsible for these schools. Anyone tracking this has a hard date on the calendar and one place to look: the annual report for fiscal 2026, which the company has filed in late September in past years.

Climb Global Solutions doesn't sell directly to end users; it sells through resellers — and a small circle of them carries increasing weight. The risk factors section of the fiscal 2025 annual report states the number plainly: "Our top five customers accounted for 55%, 54% and 51% of consolidated net sales in 2025, 2024 and 2023, respectively." The concentration isn't static — it has grown for three straight reporting years, from 51 to 55 percent.

Context matters: these are reseller partners, not end customers clustered in a single industry — if Climb lost one, another reseller could in principle serve the same underlying demand. Still, a 55 percent share resting on five business partners is a real concentration risk. Running the other way, Climb has visibly broadened its vendor base: from 22 vendors generating more than $10 million in annual sales in 2022 to 45 today, according to management on the Q2 2026 call — the reseller circle is narrowing while the vendor circle widens. The concentration has not tightened further this year: the quarterly report as of June 30, 2026 lists three major customers accounting for a combined 52 percent of net sales in the first half, versus 51 percent a year earlier.

From 2023 through 2025, Climb Global Solutions paid $0.68 per share in dividends every single year (pre-split, before the March 2026 stock split) — a reliability signal many investors read as a quality marker. The fiscal 2025 annual report (Form 10-K) ends that streak without ceremony: "Following the end of fiscal year 2025, our Board of Directors determined to suspend quarterly cash dividends on our Common Stock beginning with the first quarter of 2026 in order to preserve financial flexibility and prioritize capital allocation objectives." The freed-up capital showed up in cash almost immediately: from $36.6 million (December 31, 2025) to $56.6 million (June 30, 2026); the last $0.2 million of debt was paid off on schedule by that date, and Climb has been debt-free since.

The second half of the reversal is the more notable one. Through 2025, management consistently emphasized funding acquisitions entirely out of cash on hand on every earnings call. On the Q2 2026 call, discussing two potential large acquisitions, CEO Dale Foster said for the first time: "We're not afraid if we want to take on some debt." The $50 million credit facility with JPMorgan Chase (maturing May 18, 2028) sat undrawn as of June 30, 2026 — in the first quarter of 2026 Climb had briefly drawn $9.0 million against it for a short-term operating need and repaid it in full within the same quarter. What is new is the willingness to tap the facility for acquisitions, and it surfaced in the same window as the dividend suspension.

With fewer than seven million shares outstanding, any sizeable change in ownership stands out. The second quarter of 2026 brought two — in opposite directions. FMR LLC (Fidelity) reported 425,424 shares, or 6.1%, as of 31 March 2026. In its follow-up filing as of 30 June 2026 the figure is 194,999 shares, or 2.8% — a reduction of 230,425 shares, or 54% of the position, in a single quarter, taking it below the 5% reporting threshold.

Almost simultaneously, on 29 July 2026 and likewise as of 30 June 2026, BlackRock reported an initial stake of 419,996 shares, or 6.0%. Both filers state explicitly that they hold the shares in the ordinary course of business and without intent to influence control. The difference lies in character: a build of this size by an index provider usually follows index membership mechanically, whereas a reduction by an actively managed house is a decision. The share count barely moved over the period — so this was a change of holders, not dilution.

When Plumas acquired Cornerstone Community Bancorp in July 2025, exactly one representative of the acquired bank joined the board. The second-quarter 2025 earnings release celebrated it explicitly: the company was "thrilled to have Ken Robison, formerly a director at Cornerstone" join the boards of the holding company and the bank.

Less than eleven months later he was gone. The filing on the annual meeting of 20 May 2026 records that Kenneth E. Robison resigned "for personal reasons immediately following" the meeting — on the very day shareholders had just re-elected him. The vote itself is the second half of the finding: Robison drew 158,050 votes withheld or against, by far the highest of the entire slate; the other nine nominees ranged from 54,581 to 95,439. Less than a year after the deal, the acquired bank no longer has a voice on the acquirer's board.

In the second-quarter 2026 release the chief executive cites "deposit growth" as one of three proof points for the strength of the business model. The quarterly filing itself is considerably more precise. It states that deposits rose by $75 million, or 4%, in the first half of 2026 — "of which $41 million represents accounts that were moved from repurchase agreements to money market deposits during the current period".

Repurchase agreements are secured short-term funds that sit under borrowings on the balance sheet, not under deposits. Move the same money from the same customer from one line to the other and the deposit base grows on paper without a single new dollar entering the bank. The cross-check sits in the same balance sheet: repurchase agreements fell from $97.9 million to $59.2 million over that period. That leaves roughly $34 million of genuine new business — 1.9% of the deposit base rather than the 4% headline.

How much a US bank must set aside for expected loan losses depends on a model. One lever inside it is the "reasonable and supportable forecast period" — the span over which the bank projects the economy and derives losses from it. Plumas pulled exactly that lever in the second quarter of 2026: the quarterly filing notes in its critical accounting policies that the forecast period was extended from four quarters to eight, along with changes to certain model inputs.

The timing is what stands out. In the same six months, nonperforming loans rose from $15.1 million to $23.5 million, substandard loans from $34.2 million to $42.8 million and special mention loans from $20.1 million to $33.6 million. Yet the balance sheet allowance fell slightly, from $19.959 million to $19.740 million, and the provision charged through the income statement for the entire half-year was just $270 thousand. Coverage — allowance against defaulted loans — dropped from 132% to roughly 90%. The model change alone does not explain why the allowance failed to keep pace, but it does make the year-over-year comparison messy: the 2026 figure is produced under different rules than the 2025 one.

In its August 13, 2026 earnings release, Innovative Aerosystems bridges net income of $4,489,323 to adjusted net income of $5,979,133. By far the largest reconciling item is labelled "Amortization of acquired Intangibles" and comes to $1,299,934. The identical figure appears a few lines above in the EBITDA reconciliation as the entire "Depreciation and amortization" line — that is, depreciation and amortization combined.

The quarterly report (10-Q) filed the same day puts amortization of acquired intangible assets for that quarter at $911,678 and depreciation on property and equipment separately at $441,491. The roughly $388,000 difference is ordinary depreciation on buildings, machinery and tooling. It has nothing to do with acquisitions, yet it is added back in the adjusted measure as if it did. Over nine months the gap is wider: $3,291,520 added back against $1,980,243 of actual intangible amortization, a difference of about $1.3 million, or roughly seven cents per diluted share on adjusted earnings of $0.88.

CareCloud leases a meaningful share of its facilities from its own executive chairman: the New Jersey headquarters, guest apartments, a warehouse, the operations center in Bagh, Pakistan, and an apartment in Dubai. The rent expense is modest at $144,000 for the first half of 2026. The more interesting line sits next to it: in the first half of 2026 the company spent roughly $671,000, and in the prior-year period roughly $838,000, to upgrade "the related party leased facilities." The filing names the purpose: the 2025 spending related primarily to expanding the company's AI center.

Together roughly $1.5 million across two half years — just under 9 percent of total equity as of June 30, 2026 ($17.4 million); the first-half 2026 amount alone equals roughly a third of half-year net income ($2.0 million). Improvements to a leased building stay with the landlord when the lease ends. The same note also discloses that a company controlled by the executive chairman's son has provided artificial intelligence consulting since July 2025 for $15,000 per month ($90,000 in the first half of 2026, roughly half of it capitalized as internally developed software).

On April 14, 2026, CareCloud filed a prospectus supplement (424B5) for an at-the-market program — permission to sell new shares into the market at any time, up to $60 million in aggregate. The prospectus does the dilution math itself: at an assumed sale price of $2.70 that would be 22,222,222 new shares — against the 42,493,859 shares outstanding as of July 30, 2026, an increase of roughly 52 percent.

Nothing had been drawn through June 30, 2026: the half-year cash flow statement shows no line for proceeds from share issuance, and the share count rose by only 55,910 shares from employee equity vesting. The chief executive explicitly called the program "not as a financing we plan to lean on" on the May 7, 2026 earnings call. It stays loaded nonetheless: it is the single largest dilution lever in the capital structure, and whether it is pulled shows up in exactly one line of the next quarterly report.

CareCloud's $50 million credit facility is secured not only by company assets but also by the executive chairman's private property. On July 22, 2026, Mahmud Haq and two trusts controlled by him and by his wife pledged 4,300,000 common shares to Citizens Bank as additional collateral. On the same day the company issued him a warrant on 4,300,000 new shares at $5.00 with a five-year term (Form 10-Q filed August 6, 2026, subsequent events note, and Form 4 filed July 24, 2026).

The scale: 4.3 million shares equal roughly 10.1 percent of the 42,493,859 common shares outstanding as of July 30, 2026. The warrant cuts both ways. At $5.00 it is far out of the money and costs other shareholders nothing unless the stock roughly doubles — at which point it immediately takes about a tenth of the gain. The pledge runs the other direction: if the company falls into an uncured default, the bank may liquidate the pledged 4.3 million shares after 15 business days. Both hang on the same credit terms (Form 8-K/A filed July 24, 2026).

As of December 31, 2025, 15.0 percent of net trade receivables sat with a single customer — a year earlier, no customer exceeded 10 percent. The concentration is new, and the same half-year brought a $1.1 million write-off tied to a customer bankruptcy: the total provision for bad debts rose from $0.434 million to $1.539 million (+255 percent versus the prior-year half).

Rocky Brands reports a "net" tariff effect of roughly $15.0 million — equal to 104 percent of its own "adjusted" quarterly result of $14.4 million ($1.90 per share). The July 28, 2026 press release adjusts out only $0.7 million of acquisition-related amortization — the roughly $15.0 million tariff windfall stays fully inside the "adjusted" number.

That makes the company's own comparison metric larger than the result it is supposed to normalize: strip out the tariff effect and adjusted earnings per share would sit well below the communicated $1.90.

An $18.0 million pre-tax benefit equals 119 percent of Rocky Brands' first-half 2026 net income ($15.1 million); the receivable still open as of June 30, 2026 ($16.8 million) equals 6.4 percent of equity. A February 2026 U.S. Supreme Court ruling against invalidated IEEPA tariffs turned a cost line into a revenue line — strip it out and quarterly EBIT falls from $19.7 million to $1.7 million, below the prior-year quarter's $7.2 million.

As of June 30, 2026, only $3.7 million of the $20.5 million in tariffs paid had actually been refunded; $16.8 million sat on the balance sheet as a receivable. After the balance-sheet date, another $8.2 million was received — leaving roughly $8.6 million still outstanding.

For four years, Consolidated Water's Mexican subsidiary AdR developed a 100-million-gallon-per-day desalination plant in Baja California — until the government terminated the underlying contract outright in June 2020. The company pursued international arbitration and reached only a partial recovery in 2024: the sale of the purchased land parcel for roughly $32.0 million. The Mexican subsidiaries (CW-Cooperatief, NSC, AdR) have been under final dissolution since the first quarter of 2026 — six years after the project collapsed.

This history is already woven into the main analysis as an uncomfortable truth, because it directly informs how to read the ongoing Hawaii risk. Recorded separately here because it's worth tracking beyond the article itself: if the Hawaii project also fails or slips by more than another year, that would be the second time in six years that a Consolidated Water megaproject got stuck in a foreign or U.S. permitting chain — a pattern, not a coincidence.

The quarterly report (Form 10-Q) for the period ended June 30, 2026 contains a number that gets only a brief mention in the main analysis but deserves its own note: of the $18.8 million owed to subsidiary CW-Bahamas by the government-owned utility WSC, 64 percent is classified as "delinquent" — as of December 31, 2025, it was even higher at 71 percent. Still, Consolidated Water records no material allowance for credit losses, citing its payment history ("all previous delinquent accounts receivable from the WSC … were eventually paid in full") and an April 2026 Moody's upgrade of the Bahamas (to Ba3 from B1). The filing itself, though, states: "The delay in collecting these accounts receivable has adversely impacted the liquidity of this subsidiary."

Worth tracking going forward: the supply contracts (for example, the Blue Hills plant, running through 2032) obligate CW-Bahamas to guarantee minimum delivery volumes regardless of whether it gets paid on time. If the delinquent share climbs further in an upcoming filing, or stays elevated across several quarters instead of normalizing by year-end as it has in the past, that would signal the "the customer eventually pays" track record is running into its limits.

Since July 2025, a letter of intent has been in place to sell the operation of a geriatric behavioral hospital operator that owes Community Healthcare Trust rent and loan payments across six properties — contractually about $5.7 million a year combined ($3.2 million rent plus $2.5 million loan payments), but actually paid just $0.4 million in the second quarter of 2026. The $17.0 million plus $2.7 million loans (together $19.7 million) are fully reserved. On the call for the fourth quarter of 2025, CFO William Monroe explained why the deal has taken so long: "the buyer is still very interested in all 6 hospitals and the goal is for this transaction to happen all at one time … there would be no plans to have any sort of a staged closing" — the buyer wants all six hospitals in one transaction, not a staggered close.

That turns an ordinary troubled tenant into an all-or-nothing event: either the buyer closes the entire deal for all six properties — with new leases at a rent still unknown — or rent stays at the current $0.4 million a quarter. On the call for the second quarter of 2026, the company named a close "by year-end" 2026 as its goal, with no guarantee.

On the very same day Community Healthcare Trust cut its dividend by 31 percent, the company introduced a new, stricter metric in its press release: FAD (Funds Available for Distribution) — AFFO minus leasing commissions, tenant improvements and recurring maintenance capital spending, in other words what is actually left over after every real cost of running the portfolio. Run the old $0.48 dividend against the FAD per share the company reported for the second quarter of 2026 ($12.994 million on 27.752 million diluted shares = $0.468 per share), and the old dividend sat above FAD — at roughly 103 percent. Summed over the trailing four quarters, the FAD payout ratio comes out at around 110 percent.

Anyone who tracks this metric going forward has a clean early-warning signal: the new quarterly dividend of $0.33 equals about $9.5 million in total distributions per quarter — measured against the most recently reported FAD of $12.994 million, that would be a ratio of roughly 73 percent. Whether that improvement holds shows up only in the next quarterly report.

First National repaid $13 million of subordinated notes at par in 2025 — and, contrary to how it first looks, the two did not come from the same place. The annual report separates them cleanly: the $5 million note at 5.50 percent was issued by First National itself on June 29, 2020, and redeemed on October 1, 2025. The $8 million note at 6.00 percent, by contrast, came in with the Touchstone acquisition (closed October 1, 2024) and was redeemed on November 15, 2025. What remains is a third note — also assumed from Touchstone: originally $10 million, of which $9.5 million was still outstanding as of June 30, 2026 (the bank repurchased $500,000 of it in the second quarter of 2025), currently fixed at 4.00 percent. The annual report names the exact date that changes: starting January 30, 2027, the rate resets quarterly to the three-month SOFR (the U.S. reference rate that floating-rate loans key off) plus 596 basis points — and from that same date the bank may, for the first time, call the note with regulatory approval.

How expensive such a reset gets, the bank has already learned firsthand. On its subordinated debt, First National paid an average of 8.31 percent in 2025 (2024: 6.78 percent) — precisely because the other two notes flipped from their fixed coupons to SOFR plus a spread during the year. For the remaining note that means: substantially more expensive from January 2027 than today's 4.00 percent — exactly how much depends on where SOFR stands at the reset date and is not knowable today. The track record points toward a payoff rather than sitting on the higher rate: management justified repaying the other two notes in the annual report by saying it would "position the Company for improved profitability in future periods" — and that same logic applies even more strongly to the remaining note once January 2027 arrives. At $9.5 million against $193.6 million of equity the amount itself is small, but the interest-expense swing is not.

Between October 2021 and October 2023, First National bought loans to "health care professionals" from a third-party finance company — not at face value, but with a premium attached. As of June 30, 2026, $11.9 million of that portfolio remains on the books: $8.5 million in loan balances plus $3.4 million of unamortized premiums — the premiums alone make up roughly 40 percent of the remaining face value. The finance company used to offer buyback protection for loans and premiums when a loan stopped performing; the quarterly report states plainly, in the asset-quality section, that this protection no longer exists.

Of the $11.9 million, $1.8 million is already classified as non-accrual (of which $653,000 is the premium portion), carrying a specific reserve of $1.3 million. The premium balance stood at $4.1 million at the end of 2025 and $5.8 million at the end of 2024 — so it is already melting down, on schedule, over an average remaining term of five years. The annual report itself warns that prepayments could accelerate premium amortization and materially reduce future-period earnings. Against $193.6 million of equity, the portfolio equals roughly 6.1 percent of the bank's balance-sheet substance.

Chain Bridge Bancorp has two share classes: Class A with one vote, Class B with ten. Anyone wanting to sell a Class B share must first convert it into a Class A share — losing nine tenths of its voting power in the process. That is exactly what has been happening since the IPO: the Class B count fell from 3,512,370 (December 31, 2024) through 3,417,971 (June 30, 2025) and 3,264,680 (December 31, 2025) to 3,173,390 as of June 30, 2026. That is 338,980 shares, or 9.7 percent of the class, in eighteen months, while total shares outstanding stayed unchanged at 6,561,817.

The annual report (10-K) for 2025 spells out the consequence itself: because every conversion removes votes from the Class B pool, the relative voting power of the remaining Class B holders rises — notably that of chairman Peter G. Fitzgerald and the Fitzgerald Family, which as of December 31, 2025 held roughly 70.9 percent of Class B shares and 64.4 percent of all votes. Individual family members cashing out therefore does not dilute control, it concentrates it. For Class A holders each conversion also means extra supply in a stock that traded only about 16,000 shares on a typical day in August 2026.

As of December 31, 2025, Chain Bridge Bancorp reported a tidy deposit base in its annual report (10-K): not a single account exceeded 5 percent of total deposits. Eleven clients each held more than 1 percent and together accounted for 31.0 percent. After the mass outflow of April 15, 2025, when $506.5 million left across six accounts, that looked like a deliberately de-clustered liability side.

Six months later the de-clustering is gone. The quarterly report (10-Q) as of June 30, 2026 names three clients each above 5 percent of deposits, together $401.2 million, or 20.0 percent. That is the same structure that produced the shock in the first quarter of 2025, when three accounts held $472.0 million, or 30.1 percent. In parallel, off-balance-sheet ICS One-Way Sell deposits rose from $359.9 million (December 31, 2025) to $668.0 million; they are the buffer the bank intends to draw on if another outflow comes. Ahead of the November 2026 midterms the cluster is growing, not the spread.

Smith-Midland is a thinly traded stock with only 5.3 million shares outstanding — which makes it matter even more who holds how much of it. By far the largest shareholder is Thompson Davis & Co., Inc., a registered investment adviser based in Richmond, Virginia. In its Schedule 13D/A filed December 10, 2024, the firm reported 1,758,865 shares, or 32.86%, as of December 6, 2024 — with an entry that looks contradictory at first glance: "Sole Voting Power: 0 … Sole Dispositive Power: 1,758,865." The source of funds is listed as "OO – RIA on behalf of clients" — meaning the shares economically belong to the advisory firm's clients, even though the firm itself has full discretion over buying and selling them.

By March 31, 2026, the stake had grown further to 1,841,775 shares, or 34.71%, according to fundamental-data ownership records. The freely tradable float, per Börsenlotse/TickerGuard data, is only about 2.7 million shares — roughly half of total shares outstanding. Concentration of this size in a single investment adviser means: if that one adviser decides to recommend selling to its clients, it hits a stock that trades only a few thousand shares a day, with full force.

The article covers the short version. The 2025 proxy statement spells it out without ambiguity: "Prior to November 19, 2024, we qualified as a 'controlled company' because TRGI had the right to appoint five out of eight of the directors on our Board pursuant to our Bye-laws." TRGI is The Resource Group International Limited, a Bermuda-based investment holding company. As a "controlled company," IBEX was exempt from several Nasdaq listing requirements — including the requirement for a majority-independent board.

On November 19, 2024, IBEX repurchased 3,562,341 of its own shares directly from TRGI under a Purchase Agreement; TRGI lost its right to appoint directors, and the "controlled company" status ended that same date — though the transition exemptions did not fully phase out until June 16, 2025, per the proxy statement. What remained: TRGI still held 1,731,574 shares, or roughly 12.9 percent, as of the proxy statement dated October 28, 2025 — by far the largest single shareholder, well ahead of the entire management team combined (3.4 percent). And Chairman Mohammed Khaishgi remains TRGI's founder and CEO to this day, as well as chairman of TRG Pakistan Limited, a company the proxy statement describes as "affiliated with TRGI." A further TRGI subsidiary, TRG Holdings LLC, still leases office space from IBEX in Washington, D.C. as a "tenant at will" for $12,203 per month, per the proxy statement.

RCM Technologies' "contract assets" — revenue the company estimates it has already earned but is contractually allowed to bill customers only at a later date — rose from $10.543 million as of January 3, 2026, to $16.475 million as of July 4, 2026: a 56 percent increase in two quarters, and more than three times a single quarter's net income ($4.9 million). The quarterly report itself acknowledges that the estimate behind this requires judgment: "This approach requires significant judgment, particularly with respect to estimated total costs, project margins, and progress toward completion." In that very same period, the company discloses four auditor-identified material weaknesses, including inadequate documentation of management review processes across its business process cycles.

Rising contract assets aren't an alarm signal by themselves — they fit the low-margin EPC project work that drove most of the growth in the second quarter of 2026. The question is whether these estimated values convert into real invoices and cash receipts once the projects are billed out — and whether the acknowledged material weaknesses bite exactly at these kinds of judgment-dependent balance-sheet items.

Between August 14 and 21, 2026 — right after the August 13 quarterly report and in the middle of the stock's rally from $28.92 to over $40 — three RCM Technologies executives together sold 216,911 shares worth about $8.1 million: Executive Chairman & President Bradley Vizi sold 100,000 shares (≈$3.79 million), CFO Kevin Miller 94,413 shares (≈$3.48 million), and division head Michael Saks 22,498 shares (≈$0.84 million). For comparison: the entire net income of the quarter that triggered this rally was $4.9 million — the three insiders' sales clearly exceeded it.

Most of the sales, per the Form 4 footnotes, ran under pre-established 10b5-1 trading plans (automatic sale plans set up months or years in advance so insiders don't trade on current information) — the same plans already ran in April and May 2026 (111,711 shares ≈$3.4 million). What's new is the pace: more than double the volume in just over a quarter of the time. And one exception stands out: for Saks' sale of 10,000 shares on August 18, the Form 4 filing lists no 10b5-1 plan.

The merger agreement between Western Union and Intermex contains two break fees, and they aren't mirror images of each other. If the deal fails because Western Union can't secure antitrust clearance, Western Union owes Intermex $27.3 million. If Intermex walks away instead — say, because a third party makes a better offer and the board changes its recommendation — Intermex owes Western Union only $19.8 million (Form 10-Q for the period ended June 30, 2026). The $7.5 million gap is small next to a deal worth roughly $483 million (30.2 million shares at $16.00) — but it's a data point: a comparatively low exit cost for Intermex keeps the door for a topping bid slightly more open, economically, than a symmetric or higher fee would.

No third-party bidder has surfaced so far, and the board hasn't changed its recommendation for the Western Union deal. So today this isn't an independent signal — it's a watch item: should a new offer appear before November 10, 2026, the contractual bar for it is lower than the headline deal size would suggest.

Between fiscal year 2023 and the second quarter of fiscal year 2025, Espey received two government grants totaling $10.8 million — $7.4 million in fiscal year 2023, another $3.4 million in the second quarter of fiscal year 2025. Per the quarterly report (10-Q) as of March 31, 2026, the funds support the modernization of plant facilities and test equipment for testing and qualification for the United States Navy, and are part of the Navy's investment program "Surface Combatant Industrial Base." Measured against equity as of June 30, 2025 ($56.4 million as of March 31, 2026, $50.8 million before that), that is roughly 21 percent; measured against market capitalization (about $188 million, August 26, 2026), roughly 6 percent — a substantial sum for a company this size.

In the first nine months of fiscal year 2026, Espey invested $2,800,998 in capital equipment, of which $2,029,608 was reimbursed from the $3.4 million grant — the company's own share was about $855,000. The flip side of the opportunity: a supplier whose test equipment the Navy helps finance is more tightly bound to that one customer than an interchangeable vendor would be — and therefore more dependent on that customer's budget decisions.

On August 5, 2026, two days after filing its quarterly report, Viemed Healthcare filed a registration statement on Form S-8. It registers 913,542 additional shares for the 2024 Long Term Incentive Plan and states the running total: after this filing, 7,696,717 shares are registered for issuance under the plan. Measured against the 38,088,228 shares outstanding at June 30, 2026, that is 20.2 percent of the count. Registered is not the same as issued — actually outstanding at June 30, 2026 were 3.380 million options at a weighted average exercise price of $5.40, 2.416 million restricted stock units and 0.573 million cash-settled phantom units. Still, the registered pool shows how much room there is.

It gets interesting alongside the buyback. Viemed retired 1,976,441 shares under its 2025 program and another 680,802 in the first half of 2026 for $6.5 million, plus 272,460 shares worth $2.0 million withheld to cover payroll tax on vesting awards. The diluted share count did not move: 41,125,716 in the second quarter of 2026 against 41,083,760 a year earlier — marginally higher. The offset sits in the line "Stock options and other dilutive securities," which rose from 1,568,513 to 2,880,225. Part of that is arithmetic: under the U.S. treasury stock method, more options count automatically as the share price rises, and the closing price went from $7.43 at December 31, 2025 to $11.40 at June 30, 2026. For the shareholder the result is the same either way: two years of buybacks have not yet reduced the number of shares the earnings are divided by.

In the first quarter of 2026, 43.0 percent of BK Technologies revenue came from agencies and departments of the U.S. federal government — up from 10.9 percent in the same quarter a year earlier. The second quarter of 2026 came in at 26.9 percent, and the first half at 34.5 percent (versus 11.5 percent in the prior-year half, per the filing). Alongside that sits a single unnamed buyer the filings call only "Customer A": 39.0 percent of revenue in the first quarter of 2026, and still 14.3 percent in the second. For context, the federal share was 29 percent for full-year 2025 and 38 percent for 2024 (fiscal 2025 Form 10-K).

Visibility is shrinking at the same time: backlog fell from $21.8 million to $14.2 million as of December 31, 2025 — close to a halving in one year. Rising concentration paired with a shorter runway deserves attention when the customer in question depends on an appropriations process. The precedent is recent: during the government shutdown in the fall of 2025, federal customers asked in writing that finished shipments be held back (quarterly earnings call of November 6, 2025).

Caledonia bought the Bilboes project in January 2023 from an ownership group that had controlled it since a 2003 management buyout. The annual report (20-F) for 2025 lists Victor Gapare as a Caledonia Executive Director since 2023, noting he is a "Former chief executive officer of Bilboes Holdings (Private) Limited." Gapare is today the very executive who runs the Bilboes project at Caledonia that he previously owned and sold. He holds his Caledonia stake through Toziyana Resources Limited, a company the annual report describes as "ultimately owned by a family trust of which Mr. Gapare is the settlor" — as of April 2, 2026, Toziyana held 2,443,372 shares, or 12.66% of Caledonia's outstanding stock, the largest single shareholder block ahead of the second-largest (Shining Capital Holding II L.P., 9.96%) and BlackRock (5.54%).

This is fully disclosed and not inherently a problem — if anything, it ties Gapare's personal wealth tightly to the success of the very project he runs. But it makes any future change in this position worth watching: an insider with special knowledge of Bilboes who holds more than 12% of the company sends a signal with every purchase or sale that goes beyond the ordinary.

In January 2026, Caledonia closed an offering of $150 million of convertible senior notes (5.875% coupon, conversion price about $40.51 per share). The annual report (Form 20-F) for 2025 is precise on this point: "The Notes will mature on January 15, 2033, unless earlier converted, redeemed, or repurchased." The later interim report (Form 6-K) for the first half of 2026, dated August 10, 2026, contradicts itself, however — the "Liquidity and Capital Resources" section refers to "Convertible Senior Notes due 2030," while the "Financing activities" section of the same document correctly says "due 2033." For a bond worth roughly 31% of the company's market capitalization (about $489 million as of 08/25/2026), that is not a cosmetic error — it is evidence of how much of the investor relations team's bandwidth is currently going toward the Bilboes growth project rather than proofreading its own disclosures.

For shareholders, the dilution mechanism matters more: if the share price rises above the $40.51 conversion price, noteholders can convert into shares; $14.4 million of purchased capped call options raise the effective dilution threshold to $56.72. Both thresholds sit well above the current share price, but they are the yardstick against which any future dilution will be measured.

On July 1, 2026, ReposiTrak bought 4,709,837 more shares of SPAR Group, Inc. for roughly $3.3 million — partly financed through an unsecured promissory note at 6 percent interest. On July 20, 2026, ReposiTrak disclosed via Schedule 13D that, combined with shares received earlier as payment, it now held 8,900,406 shares, or 31.4 percent, of SPAR Group — with the explicit option to hold "discussions with the board of directors" or evaluate "strategic alternatives." What the filing does not mention: SPAR Group already carried two publicly disclosed Nasdaq deficiency notices at that point — from January 12, 2026 (share price below $1, disclosed before all three ReposiTrak capital moves) and from April 8, 2026 (equity below the $2.5 million minimum, disclosed after the March 17 loan but before the May 29 stock election and the July 1 share purchase) — both publicly disclosed via SPAR Group's own Form 8-K filings. Only on July 14, 2026, six days before ReposiTrak's 13D filing, did Nasdaq tell SPAR Group its stock — trading below $1 since December 2025 and with stockholders' equity below the $2.5 million minimum required for continued listing — would be delisted from the Nasdaq Capital Market at the open of trading on July 23, 2026, three days after ReposiTrak's filing, not before it. SPAR Group has traded on the OTCQB market, not a national exchange, ever since.

SPAR Group's own Form 10-Q for the quarter ended June 30, 2026 shows net revenue down year over year ($36.9 million versus $38.6 million) and only $2.9 million in cash. On top of the equity stake, ReposiTrak had also extended up to $4.0 million in financing (of which $3.0 million was already advanced, at 8 percent interest, as of March 31, 2026) to SPAR Group subsidiary SPAR Marketing Force, Inc. Altogether, more than $8 million of ReposiTrak's capital — against equity of $50.5 million as of March 31, 2026 — now sits with a single customer that was pushed off Nasdaq.

The stock consideration in the Worldpay deal has a side effect that never shows up on a scanner: 43,268,041 newly issued Global Payments shares went to GTCR LLC and other former Worldpay owners outside FIS on January 9, 2026. According to fundamental data (as of March 31, 2026), GTCR LLC held roughly 15.82 percent of all outstanding Global Payments shares as a result — more than any other single shareholder, ahead of BlackRock (6.91 percent), Vanguard Capital Management (5.40 percent) and Harris Associates (4.33 percent). GTCR is a Chicago-based private equity firm; private equity investors typically do not hold stakes indefinitely, instead winding them down in an orderly way once lock-ups and market windows allow it.

For shareholders that is a classic "overhang" risk: a single large holder with roughly 43 million shares can put noticeable pressure on the stock if it sells, regardless of how the underlying business performs. The SEC filing on deal closing explicitly labels the stock consideration an "unregistered sale of equity securities" (Item 3.02) — a designation that triggers its own disclosure requirements once these shares are sold. Anyone tracking the position will find the next development in GTCR's routine SEC filings.

On May 11, 2026, International Seaways terminated an existing equity distribution agreement dating from December 2023 and signed a new one the same day. Through four banks, the company may sell common stock worth up to $200 million continuously over the exchange ("at the market"). The quarterly report (10-Q) for the period ended June 30, 2026 records that, as of the filing date of August 10, 2026, not a single share had been sold under it.

The size is material: $200 million equals roughly 8.8 percent of the $2,265.3 million of equity (as of June 30, 2026), close to half of cash plus short-term investments ($409 million), and at $99.52 a share about 2.0 million shares, or a good 4 percent of the 49,531,311 shares outstanding as of August 6, 2026. The timing is what stands out: in that same quarter the company generated $260.7 million of free cash flow, held roughly $935 million of total liquidity and reported a net loan-to-value of about 6 percent — while the $50 million repurchase program runs out unused at the end of 2026.

Anyone tracking dilution will find the answer in the next quarterly report: the diluted share count for the quarter ended June 30, 2026 was 49,857,565. A clear increase means the window has been used.

Look for Pampa Energía's largest single shareholder and you will not land on Chairman Marcos Marcelo Mindlin (13.87 percent) or the entire management control group (roughly 21.85 percent combined) — you will land on the Argentine state. The ownership table in the annual report (20-F) for 2025 shows ANSES, the national social security agency, with 311,029,993 shares and 22.81 percent of both capital stock and voting power — more than any private major shareholder. The footnote explains where it came from: in 2008 Argentina nationalized the private pension funds and transferred their share holdings, originally including 295,765,953 Pampa shares (20.50 percent), to the newly created ANSES. Since a 2011 decree, ANSES may exercise its voting power in proportion to its actual stake — without the 5 percent cap that used to apply to private pension funds.

For investors that is not a footnote: nearly a quarter of the company sits with an agency whose voting behavior is shaped by whichever government is in office — and whose position could change through sales, a partial privatization, or political direction, without any change to Pampa's own operations. A retreat by ANSES from the position (or, conversely, an increase) would be a signal that shows up in no revenue or earnings figure, yet can move the stock.

Read only the headline "record revenue, guidance raised twice" and you miss a number buried inside ITT's own earnings release: reported (GAAP) earnings per share for 2026 are guided to land between $4.47 and $4.67 — at the midpoint roughly 25 percent below the $6.11 ITT actually earned in 2025. The reason is the cost of the SPX FLOW acquisition: in the second quarter of 2026 alone, scheduled amortization of acquired customer relationships and technology jumped from $11.6 million to $62.6 million, on top of one-time deal costs ($67.5 million in the first quarter of 2026 alone). Adjusted earnings per share, which strips out these items, is guided to rise 13 to 16 percent instead — the gap between the two metrics has itself become a number worth watching.

This is not automatically a red flag: amortization and one-time costs in the first year of a billion-dollar acquisition are normal for an industrial company, and the operating business is running ahead of plan by management's own account. But it shows how much a single deal can distort the income statement of a company whose own per-share profit already dipped slightly in 2025 (to $6.11). Whether the gap between GAAP and adjusted earnings narrows as one-time costs roll off — or stays structural because the SPX FLOW intangible amortization runs for years — is something only the coming quarterly reports can show.

Dycom's annual report (10-K) for fiscal 2026 (filed March 9, 2026) lists three customers that each exceeded 10 percent of contract revenues: AT&T Inc. at 25.4 percent ($1,410.5 million), Verizon Communications at 14.0 percent ($778.1 million) and Lumen Technologies at 10.8 percent ($598.6 million). Combined, more than half of total contract revenue came from three telecom carriers. AT&T's trajectory is the sharpest: 16.9 percent in fiscal 2024, 20.1 percent in fiscal 2025, 25.4 percent in fiscal 2026 — a roughly 8.5-percentage-point rise in two years, while total revenue grew only about 33 percent over the same span.

The trend could continue: on February 2, 2026 — after fiscal 2026 closed — AT&T completed its acquisition of Lumen's mass-markets fiber business. Dycom's 10-Q still reports revenue from that business under Lumen for now, but once the reporting catches up with the ownership change, AT&T's disclosed share is likely to climb further while Lumen's falls. For a contractor whose largest single customer keeps gaining weight, the next customer-concentration table in the quarterly report is not a footnote — it is one of the most important numbers in the filing.

In September 2025, Avnet issued convertible senior notes due 2030 for $650 million at a coupon of just 1.75 percent — cheap financing, as long as nobody converts. The initial conversion price is roughly $70.27 per share. Noteholders may convert early once Avnet stock trades above roughly $91 for at least 20 of 30 consecutive trading days (fiscal 2026 annual report 10-K, long-term debt note). On August 6, 2026, the stock touched $100.00, its first close above that threshold, but fell back to $88.64 by August 21, 2026 — the 20-of-30-day condition has, by our research, not yet been met.

At a $70.27 conversion price, $650 million in principal works out to roughly 9.25 million additional shares — about 11 percent of the 82,082,259 shares outstanding as of June 27, 2026. On conversion, Avnet must settle the principal amount in cash and may settle anything above that in cash, stock, or a combination, at its discretion. Starting September 8, 2028, Avnet itself may call the notes if the same price threshold is met. Anyone tracking dilution risk in this stock should watch not just the price, but the 20-of-30-day count itself.

On February 27, 2026, 4iG Plc signed a loan agreement with Mubadala Investment Company PJSC, the sovereign wealth fund of Abu Dhabi. The Q1 2026 flash report describes it in its executive summary: a loan worth 50 million U.S. dollars with a three-year term that will be mandatorily converted into shares at the end of the term — not at the lender's option, but by obligation. Conversion falls due in 2029.

At the exchange rate of August 24, 2026 (1 U.S. dollar = 310.06 forint), 50 million dollars equal roughly HUF 15.5 billion. Measured against the HUF 124.7 billion of equity attributable to 4iG shareholders as of March 31, 2026, that is a good 12 percent; measured against the HUF 118.4 billion cash balance, roughly 13 percent. Free float on the same date was 10.11 percent. How much the conversion dilutes depends on the conversion price, which the company has not published — the lower the share price in 2029, the more shares are created. That condition is the open question.

In its capital structure table, the 2025 annual report puts net debt at HUF 917,118 million against EBITDA of HUF 293,732 million — a ratio of 3.12x, down from 3.88x a year earlier. Note 39 of the same report states that the company has met its financial commitment with that figure. The commitment itself sits a few pages earlier: for the loans of 4iG Távközlési Holding Zrt., net debt to EBITDA may not exceed 4.0x from January 1, 2023, and the debt service coverage ratio may not fall below 1.2 from January 1, 2024. Testing is annual, with a deadline of June 30 of the following year, based on IFRS accounts prepared by an international audit firm.

Three months later the arithmetic looked different. The Q1 2026 investor presentation puts net debt as of March 31, 2026 at HUF 1,068,016 million and the ratio at 3.6x. The cushion to the cap has fallen from 0.88 to roughly 0.4 in one quarter, driven by the defence acquisitions closed in that quarter and by a 176.6 million euro bond. Anyone tracking the number will find it on the same slide of every quarterly presentation.

On June 16, 2026 — exactly 25 days after China's securities regulator imposed roughly RMB 411 million of fines and disgorgement on UP Fintech subsidiaries — an amended beneficial ownership report (Schedule 13G/A) reached the U.S. securities regulator, the SEC. The filers: Avenir Tech Limited and the entities behind it, Avenir View Limited and Avenir Investment Holdings Limited, all registered in the British Virgin Islands and ultimately attributable to Mr. Lin Li. Reported holding: 19,487,703 ADSs, equal to 292,315,545 Class A ordinary shares and 10.9 percent of the Class A and Class B ordinary shares outstanding as of March 31, 2026.

The comparison is what makes the filing interesting. In the annual report on Form 20-F for 2025 (filed April 24, 2026) UP Fintech still lists the same group at 160,013,700 Class A ordinary shares, or 5.82 percent, based on a Schedule 13G filed April 29, 2025. Between the two filings the reported position has therefore almost doubled. That makes Avenir by far the largest shareholder outside the founder circle — roughly ten times the size of what director Liu Jian sold on June 25, 2026 after filing a Form 144. Anyone tracking the ownership side of this stock should watch for further amendments: an increase beyond 10 percent, or a switch from a passive Schedule 13G to an active Schedule 13D, would be the first sign of a shareholder who wants a say.

The large holding disclosure of August 14, 2026 — one day after the half-year report — lists twelve contracts covering the shares held. Each carries the same label: 주식담보대출, a securities loan against pledged shares. Total: 5,922,914 shares, or 25.33 percent per the filing — roughly three quarters of the 33.43 percent block held by Ocean B Holdings and its related parties. The loans run with eight lenders and total about KRW 32.1 billion at interest rates of 5.04 to 6.90 percent.

The trigger sits in the column headed 담보유지비율: maintenance ratios of 110 to 180 percent. If the price falls far enough that the pledged shares slip below that value, the lender can demand more collateral or sell — additional selling pressure exactly in a downturn. The frequency shows how tight it is: five contract amendments between June 26 and August 14, 2026, with the shortest terms expiring in October and November 2026.

On February 13, 2026, MK Electron reported a preliminary 2025 group net profit of KRW 21,575 million ("return to profit"). The audited annual report filed March 23, 2026 shows KRW 13,765 million for the same year — 36 percent less, a good five weeks later. Revenue was practically unchanged (KRW 1,404.3 versus 1,403.8 billion); the gap arose below the operating line.

For investors this means: at this company, the preliminary February numbers are a rough sketch, not an audit. Anyone trading the year-end earnings flash should price in the March correction — in a group full of real estate trust positions, fund vehicles and metal derivatives, there is a lot of room between preliminary and audited.

The segment note of the half-year report filed August 13, 2026 lists a domestic key customer "Company A" with KRW 145,291 million in H1 2026 revenue. The comparison column for H1 2025 shows: 0. A single new buyer thus accounts for almost 16 percent of group revenue of KRW 915,448 million — and roughly 43 percent of the entire product revenue increase of KRW 341 billion over the prior-year half.

The report does not name the customer. What is certain: the half-year growth that carried the stock in the spring of 2026 hangs to a considerable degree on this one relationship. Whether it holds in the second half will show in the key customer table of the next report.

In mid-2026 (secondary sources differ between June 15 and July 15, 2026 for the date), Chairman Lee Hee-hwa transferred 3,500,000 common shares at no cost -- about 97 percent of his direct shareholding, or roughly 12.8 percent of the entire company -- to his son, Vice President Lee Jong-myung. The son's stake rose from 1,993,281 to 5,493,281 shares (about 20.04 percent); Lee Hee-hwa himself kept only 94,335 shares (0.34 percent). The family's combined stake stayed unchanged at about 45.7 percent -- this was a pure intra-family transfer; the transfer was filed with DART on July 15, 2026.

The timing stands out: per an assessment by etoday.co.kr (article dated August 7, 2026), market capitalization stood at only about KRW 51.1 billion -- about 12 percent of 2025 annual revenue of KRW 426.2 billion. The same business outlet called the transfer a "low-price succession" (저가 승계) in its headline: the lower the share price at the time of a gift, the lower the gift tax owed under Korean law. Whether the timing was deliberate is not established. For minority shareholders the practical outcome is that Lee Jong-myung has become the "de facto controlling shareholder," with no change to free float or the family's overall control.

On January 5, 2026, Guyoung Technology filed a debt guarantee for its U.S. subsidiary CAR TECH, LLC in Opelika, Alabama, acquired in October 2025: the company guarantees KRW 43.263 billion (roughly $30 million) of a Korea Eximbank loan taken out by CAR TECH, term January 6, 2026 to January 6, 2027, to refinance a prior $10 million loan and draw new credit. The same filing put the group's total guarantee obligations at KRW 234.485 billion -- about 152 percent of total shareholders equity of KRW 154.0 billion as of December 31, 2025.

Only six months after the purchase, a reinvestment of roughly KRW 29.5 billion in two tranches had to follow -- about three times the original purchase price of roughly KRW 8.3 billion for the majority stake. A trade article (ibtomato.com, April 13, 2026) summed it up with the image of a "bottomless jar." Should CAR TECH need further capital, or should the Eximbank loan turn nonperforming, the guarantee would hit a sum larger than everything shareholders currently own.

The 20-F annual report for 2025 names a customer concentration the article body only touches on: "Overseas license fees and royalty payments generated from our mobile games represented 8.9% of our total revenues in 2025 [...] with 81.7% of our 2025 revenues from mobile game license fees and royalty payments attributable to license arrangements with Nuverse (Hong Kong) Limited." In plain terms: 8.9 percent of total group revenue in 2025 came from overseas-licensed mobile-game royalties, and 81.7 percent of that — a single licensee, Nuverse (Hong Kong) Limited, responsible for marketing Ragnarok X: Next Generation in Taiwan, Hong Kong, Macau and Southeast Asia. Arithmetically, that puts roughly 7.3 percent of total 2025 group revenue behind one counterparty — comfortably above the 5 percent materiality threshold.

Per the filing, the license runs for an initial two years from the start of Ragnarok X: Next Generation's commercialization in those markets, with automatic one-year extensions after that — a renewal window that recurs every year and whose outcome is not fully in Gravity's own hands.

The 20-F annual report for 2025, filed April 24, 2026, is unambiguous: "Since our inception, we have not declared or paid any dividends on our common shares. [...] We have no intention to pay dividends in the near future." On August 7, 2026 — less than four months later — the same board approved, via a 6-K titled "Resolution on Cash Dividend Payment," the first dividend in the company's 26-year history: KRW 4,400 per share, record date June 30, 2026, payment date September 2, 2026, total amount KRW 30,575,160,000 (roughly $21.7 million at that day's exchange rate).

Profit attributable to the parent company's shareholders in 2025 was KRW 67.5 billion; measured against that, the payout equals a ratio of roughly 45.3 percent — not a symbolic amount. The reversal is neither illegal nor unusual (capital returns are often decided on short notice), but it shows how little a "no intention" clause in an annual report is worth once the capital position — here, KRW 203.6 billion in cash against essentially no debt — changes quickly. Whether the interim dividend becomes a lasting policy will only be clear from the next resolution.

On August 6, 2026, HJ Shipbuilding was named preferred bidder for the Busan daily Kookje Shinmun — a roughly 10-billion-won purchase, equal to about 6 percent of the group's most recently reported cash balance. There is no business link to shipbuilding or construction; the company cites its role as the largest employer in the Busan region as the rationale.

The agreement is signed, and closing is expected after a creditors' meeting in September 2026. For a company running roughly 266 percent leverage by calculation, any off-industry capital commitment is worth watching.

The 7,028,394 new shares from the October 2025 capital raise — about 7.8 percent of today's share count — sit entirely with the already-dominant special-purpose vehicle Ecoprime Marine Pacific. They carry a one-year lock-up from their October 28, 2025 listing date, which therefore expires on October 28, 2026.

Whether the majority shareholder sells any of that block afterward is an open question — but it is the first time since the issue that a large share block would, in principle, become freely tradeable. Watch major-shareholder filings around that date.

Korea Zinc raised capital via a third-party share allotment on September 13, 2023 — which, on Youngpoong's account, diluted its own stake by roughly 5 percentage points. Youngpoong sued to invalidate that capital increase; on April 25, 2025, plaintiff status passed to its wholly owned subsidiary YPC Limited (유)와이피씨, into which Youngpoong had contributed its Korea Zinc shares. YPC won at first instance on June 27, 2025 — Korea Zinc has appealed, and the case is ongoing.

The semi-annual report spells out the consequence about as plainly as a mandatory filing ever does: if YPC's win is finalized, "the equity stake of the parent company and YPC Limited in Korea Zinc will increase" — though the outcome cannot be predicted. As of June 30, 2026, YPC directly holds 25.21 percent of Korea Zinc, with a 30.36 percent equity-accounted look-through interest including other related parties — and that stake, carried at 3,345.7 billion won, is worth nearly five times Youngpoong's own market value. Every percentage point up or down moves Youngpoong's balance sheet more than its entire operating business.

Youngpoong sued the Securities and Futures Commission (SFC) on July 6, 2026 at the Seoul Administrative Court over its June 22, 2026 accounting sanction — and simultaneously requested a stay of enforcement. The court granted that request on August 5, 2026: the corrective order, the three-year mandatory external-auditor designation, and the sanctions against current executives are suspended until 30 days after the first-instance ruling in the cancellation suit (the former CEO's personal sanction was denied a stay and remains in force). Because the stay applies, Youngpoong did not retroactively restate its first-half 2026 financial statements.

The semi-annual report itself states the number that matters: if Youngpoong ultimately loses the case, non-current provisions rise by 233.1 billion won, deferred tax liabilities fall by 56.4 billion won, and retained earnings ultimately drop by 176.7 billion won — against a market capitalization of roughly 686.7 billion won, that is not a footnote, it is about a quarter of the entire market value. The SFC itself characterized the violations as "intentional," not a defensible estimation difference — a poor starting point procedurally for Youngpoong's suit, though not necessarily determinative of the outcome.

Two headlines about Korea Aerospace Industries that ran at the same time in August 2026 share one name: Hanwha. On one side, Hanwha Aerospace builds and delivers the engines (a licensed version of the French Safran Arriel 2L2) for the LAH "Mir-On" attack helicopter — the very component whose diffuser showed corrosion in 47 of 57 delivered engines (82.5 percent) and cracks in 38 (66.7 percent), grounding the entire roughly 5 trillion won program in May 2026 (publicly disclosed June 13-14, 2026; Newspim, June 14, 2026). On the other side, that same Hanwha Group has been buying its way into KAI since late 2025: from 3.2 percent to 15.89 percent by August 10, 2026, with a further pledge to buy up to 500 billion won more in shares by year-end.

Whether and how much this overlap — supplier of a defective part and simultaneously the customer's largest private shareholder — affects the ongoing Defense Acquisition Program Administration (DAPA) investigation is not publicly documented; the agency has only announced a full recall and detailed analysis. For shareholders, the setup is still notable: KAI's possible future controlling shareholder is also the supplier whose component failure contributed to the second-quarter 2026 earnings miss.

Item 3 on the agenda of the annual meeting on September 25, 2026 is authority for a share consolidation of one-for-five to one-for-fifteen. The board may pick the exact ratio at its sole discretion, the authority runs until September 25, 2027, and fractional shares are cancelled for no consideration. Against 423.0 million common shares and 26.3 million exchangeable shares outstanding as of August 5, 2026, the outer ratio would erase roughly 93 percent of the share count — while changing the value of the company by nothing at all.

The reason is stated plainly in the proxy statement: Nasdaq requires a minimum closing bid price of $1.00. As of the date of that document, August 7, 2026, the closing price was $0.97 and had been below the threshold for 25 consecutive business days; a deficiency letter follows after 30. The repetition is what stands out: shareholders had already approved a substantially identical authorization at the 2025 annual meeting, the board never used it, and it is now being renewed. Holders should treat the gap to the one-dollar mark and the September 25, 2026 vote as an event in its own right.

The segment note of the half-year report shows a pharmaceutical manufacturing segment separately for the first time — the contract manufacturing business of Lotte Biologics, previously reported under "other". The figures explain why a separate line became necessary: revenue collapsed from KRW 88,082 million to KRW 23,889 million, down 72.9 percent against the first half of 2025. The operating loss quadrupled from KRW 29,453 million to KRW 121,302 million, and the net loss stands at KRW 127,932 million.

For scale: the entire group operating profit for the same half-year was KRW 175,055 million. The pharmaceutical segment alone therefore consumed about 69 percent of it. At the same time it ties up KRW 2,125,145 million in assets — nine percent of the group balance sheet total of KRW 23,652,508 million — and carries KRW 1,180,062 million in liabilities. The IR presentation of May 27, 2026 puts commercial production at the Songdo plant in 2027. Until then the segment is a pure drag whose loss has most recently quadrupled from one half-year to the next.

As of June 30, 2026 the holding company held 23,613,015 of its own common shares out of 99,663,776 issued — 23.69 percent. They date back to the holding company restructuring of 2017 and 2018. What is to happen to them is not answered in the half-year report's section on share capital: the company says it is reviewing ways to use them "to enhance shareholder value and improve the financial structure" and will disclose a concrete approach once one is settled. No further short-term purchases, disposals or cancellations are planned. The only substantive reason given for holding on appears as a table entry: preparation for uncertainty in the business environment.

That open-endedness now has an expiry date. The third amendment to Korea's Commercial Act, promulgated on March 6, 2026, takes effect in its main provisions on September 10, 2026. It requires companies to cancel treasury shares within one year of acquisition, with a six-month transition period for holdings acquired before the law takes effect. A company that wants to keep or sell them instead needs a retention plan approved by the board and endorsed by the annual general meeting every year; any disposal must be made pro rata to all shareholders. For Lotte that means a decision on almost a quarter of its share capital has to be on the table by the annual meeting in March 2027. Cancellation would cut the number of issued shares by almost a quarter and take the overhang — potential future selling pressure — out of the market. It would not change earnings per share, because treasury shares do not count there in the first place. What it would raise is the control stake of the largest shareholder group, which stood at 45.74 percent on March 31, 2026, without anyone paying a won for it.

Note 12 of the half-year report filed on August 14, 2026 lists the equity-accounted investees one by one. As of June 30, 2026 the two largest carry book values of KRW 3,425,490 million (Lotte Chemical, 25.31 percent) and KRW 1,886,834 million (Lotte Shopping, 40.00 percent), together KRW 5,312,324 million. Valuing the same stakes at the closing prices of August 21, 2026 gives roughly KRW 588 billion for the Lotte Chemical holding and roughly KRW 1,094 billion for the Lotte Shopping holding — about KRW 1,682 billion in total. The carrying amount is therefore some KRW 3.6 trillion, or roughly three times, above what the market pays for the same shares. For scale: total equity attributable to the holding company's shareholders is KRW 6,395,557 million.

This is not an accounting error. Under K-IFRS an equity-accounted carrying amount does not have to be written down to the share price; what matters is the recoverable amount, which includes value in use. It is, however, a documented impairment risk, and the company is already acting on it: it recognized an impairment of KRW 27,811 million on Lotte Global Logistics in the first half of 2026, after KRW 54,698 million in full-year 2025. Anyone reading the next filings should follow the impairment column in Note 12 — that is where the gap will either close or be booked.

On February 26, 2026, Opera’s board authorized a share repurchase program of up to $300 million over two years — roughly 18 percent of the $1.70 billion market capitalization as of the August 23, 2026 data date. A buyback normally shrinks the share count and lifts every remaining holder’s stake. The annual report on Form 20-F for 2025 describes a twist: the company will buy “both ADSs from the open market and pro-rata purchases of ordinary shares from our majority shareholder” — that is, stock from the exchange and shares bought directly from Kunlun Tech, which holds 68.0 percent.

In practice, Kunlun’s stake stays roughly constant through every repurchase, and part of the money spent flows to the controlling owner rather than into the market. The interim report on Form 6-K of August 19, 2026 gives the running total: through June 30, 2026, Opera had repurchased 1,776,194 shares for $28.1 million, an average of $15.79 per share. In the second quarter alone it repurchased 0.64 million shares for $11.1 million while paying out $14.2 million in cash, because a $4.1 million commitment to the majority shareholder was settled at the same time. Read the buyback headlines as pure share-count math and you miss that second direction of payment.

Item 6.B of the Form 20-F for 2025 contains a single sentence that opens up a scale: "During the year ended December 31, 2025, we have paid aggregate cash compensation of $7.8 million to our directors and executive officers (five persons)." Five people — the board and the executive team together — received $7.8 million in cash in 2025. The company earned $12.3 million that same year. That is roughly 64 percent of net income for five individuals, at a company with about 18 shore-based employees and a general and administrative expense line of $28.1 million.

The second half of the finding sits right next to it. The audit committee consists of a single independent director, in the words of the same report: "Our Board of Directors has established an Audit Committee, consisting of a single independent director, Ms. Chu." As a foreign private issuer the company is expressly permitted to opt out of many of the corporate governance requirements the New York Stock Exchange imposes on U.S. issuers. On top of that sits a 2020 agreement under which founder and chief executive Herbjørn Hansson holds his present position until 2027 and may then become "non-executive Chairman as long as he lives." For a minority shareholder the question is not whether the numbers are right — they are audited — but who challenges them if they are not.

As of February 3, 2026 there were 1,074,872,000 A shares and 3,390,128,000 B shares outstanding, according to the annual report on Form 20-F for 2025. Every A share is held by Novo Holdings A/S, the investment arm of the Novo Nordisk Foundation — that is 76.02 percent of all voting rights on roughly a quarter of the capital; a further 177,560,500 B shares add another 1.26 percent of the votes. The filing states that these A shares cannot be sold for as long as the foundation exists, and that the foundation is required by its own statutes to maintain material influence over Novo Nordisk. The chair of the foundation also chairs the Novo Nordisk board, and the chief executive of Novo Holdings sits on it too.

The practical consequence is spelled out in Item 16G: because Novo Nordisk qualifies as a controlled company, it is not obliged to comply with three rules of the NYSE Listed Company Manual — a majority of independent directors (303A.01), a nominating committee (303A.04) and a compensation committee (303A.05). That this power is real became visible in the autumn of 2025, when an extraordinary general meeting on November 14, 2025 replaced the board. Whoever buys ADRs on the NYSE buys economic participation; a say in the company is not part of the package.

In the "Key modelling considerations" section of the first-half 2026 report (Form 6-K filed August 4, 2026) sits a sentence that is easy to skim past. Effective January 1, 2027, Novo Nordisk lowers the US list price, or wholesale acquisition cost, of Wegovy injection 2.4 mg and 7.2 mg and tablets up to 25 mg, and of Ozempic injection 0.5 mg, 1 mg and 2 mg plus Ozempic pill 7 mg and 14 mg, to a uniform $675. The company puts that at a cut of roughly 50 percent for Wegovy and roughly 35 percent for Ozempic — across every dose of those medicines.

The filing also says what it means: "The change in list prices is expected to impact Novo Nordisk's cash flow in 2027." For scale: US Operations contributed DKK 76.3 billion of adjusted sales in the first half of 2026, a little over half of group revenue. To be fair, the list price is not the realized price — rebates to payers and intermediaries sit in between, and the self-pay price for the Wegovy pill already runs at $149 to $299 per month. But an officially announced halving of the list price in the most important market is a fixed date on the calendar, not a rumor — and it is not yet reflected in the 2026 guidance at all.

In 2024 Novo Nordisk reported net profit of DKK 100,988 million — and free cash flow of minus DKK 14,707 million. A company that earns roughly DKK 101 billion while spending nearly DKK 15 billion more than it takes in is not an everyday sight. Both figures sit side by side in the Form 6-K carrying the full-year 2025 results, filed February 3, 2026; in 2025 free cash flow swung to plus DKK 28,295 million on capital expenditure of DKK 60,140 million.

The explanation is in the 20-F for 2025 under related party transactions: in 2024 Novo Nordisk bought three fill-finish sites from Novo Holdings A/S — its own controlling shareholder — in connection with Novo Holdings' acquisition of Catalent, Inc. The filing puts the purchase price for the three sites at $11.7 billion, "mainly debt-financed." As of the end of 2025 the closing mechanism adjustments with Novo Holdings had still not been finally confirmed. For investors this is notable twice over: the largest acquisition of recent corporate history ran with the counterparty that controls 76.02 percent of the votes through the A shares — and it is the reason the resulting depreciation has weighed on the income statement ever since.

The current report of July 20, 2026 contains a formula that rarely makes it into press coverage. If the relevant Tempus price is above $48.42, the exchange ratio equals "$16.25 divided by the Parent Stock Price" — meaning the value per Personalis share is always exactly $16.25, no matter how far Tempus rises. If the price is at or below $48.42, the ratio is fixed at 0.3356 and the value falls one for one with the Tempus price. At $40 that would leave $13.42 per share.

What counts is not the price on the closing date but the volume-weighted average across the fifteen trading days before closing. Personalis may terminate only if that figure falls below a second threshold of $46.00 — and under the agreement that right is expected to be exercisable solely within a two-business-day window immediately before the scheduled closing. On August 21, 2026 the Personalis share closed at $18.38, roughly 13 percent above the contractual maximum.

The concentration table in the quarterly report for June 30, 2026 documents a shift that took just twelve months: Merck & Co. accounted for 37 percent of revenue in the second quarter of 2026, up from 11 percent a year earlier. Receivables are more concentrated still — as of June 30, 2026, 49 percent of all outstanding invoices were owed by Merck, against 12 percent at December 31, 2025. Measured against half-year revenue of $37.8 million, Merck's 30 percent share sits far above any materiality threshold.

The point is the dual role. Merck Sharp & Dohme LLC holds more than 10 percent of Personalis stock and is treated as a related party in the filings — 14,044,943 shares at $3.56 each, $50 million in total. Alongside the merger agreement, Merck signed a voting agreement covering roughly 13 percent of the voting power: support for the Tempus deal, opposition to any competing proposal. The largest customer therefore helps decide on the sale of its own supplier.

Note 8 of the quarterly report for June 30, 2026 explains how Tempus AI came by its Personalis stake in the first place: 9,218,800 shares from two warrants at an average exercise price of $2.00 — and Tempus did not pay cash for those warrants. They were issued in November 2023 as consideration for its obligations under the commercialization agreement. Another 3,500,000 shares at $5.07 followed in August 2024. Against 106.8 million shares outstanding (as of July 29, 2026) that is roughly 12 percent of the capital, well above the 5 percent materiality threshold.

The same Tempus agreement expressly grants the partner the right to use the genomic data derived from the tests; if that data is licensed on to a third party, Personalis is entitled to 10 to 20 percent of gross revenue. What that arrangement earned Personalis in the first half of 2026, as in the prior-year period, was zero. A party that controls distribution, uses the data and sits at the table as a major shareholder holds an information advantage over any outsider when the price for the whole company is set. The agreement does permit a superior competing proposal — which would trigger a termination fee of roughly $76.8 million.

The notes to the 10-Q for the period ended June 30, 2026 carry a number that the earnings release leaves out. On June 28, 2026 Zymeworks signed a merger agreement with Theravance Biopharma — $17.00 in cash per ordinary share plus one contingent value right tied to future proceeds from the drug candidate ampreloxetine. If the transaction is not consummated by December 28, 2026 (a deadline that extends automatically for two three-month periods when only antitrust clearance is outstanding), either side may terminate. And for one specific case a payment is agreed: a reverse termination fee of $32.515 million payable by Zymeworks to Theravance, including where the merger fails on conditions related to clearance under the Hart-Scott-Rodino Act.

For scale: $32.5 million equals roughly 40 percent of the entire stockholders' equity Zymeworks reported on June 30, 2026 ($80.4 million). The exposure is not a price risk but a balance sheet risk — it lands without anything of value arriving in return. Such an outcome would be announced in a current report on Form 8-K.

The balance sheet in the 10-Q for the period ended June 30, 2026 shows a drop that looks like a disaster: stockholders' equity fell from $268.5 million on December 31, 2025 to $80.4 million on June 30, 2026, down 70 percent in two quarters. The accumulated deficit rose from $953.2 million to $1,171.2 million over the same period. Part of that is the $89.2 million half-year loss. The larger part is deliberate: at Zymeworks share repurchases are charged directly against the accumulated deficit, and the cash flow statement shows $128.1 million spent on buybacks in the first half of 2026.

Since the first program started in August 2024 the company has used a cumulative $213.6 million to reacquire 10,571,316 shares at an average price of $20.21 (as of August 4, 2026). The share count fell from 74,638,413 (December 31, 2025) to 71,412,072 (June 30, 2026) and stood at roughly 71.0 million on August 4, 2026. Of the 2026 program's $125.0 million authorization, $49.4 million had been used by that date, leaving $75.6 million of remaining capacity. The sequence is what deserves attention: the money for those buybacks came largely from a loan carrying a 10.7 percent effective rate.

On March 2, 2026 the subsidiary Zymeworks BC sold 30 percent of future Ziihera royalties under its agreements with Jazz Pharmaceuticals and BeOne Medicines to a newly formed special purpose entity. That entity borrowed $250.0 million from Royalty Pharma on a secured, non-recourse basis; after $5.563 million of transaction costs the cash flow statement shows $244.7 million in net proceeds. The 10-Q for the period ended June 30, 2026 puts a number on the repayment: roughly $481.3 million by the December 31, 2042 maturity date, or about $412.5 million if the loan is repaid in full by the end of 2033 — in each case inclusive of interest, yield protection premiums and exit fees.

The filing states an estimated effective interest rate of 10.7 percent as of June 30, 2026; interest expense in the second quarter of 2026 alone was $6.6 million. Until the loan is repaid, 30 percent of Ziihera royalties flow to Royalty Pharma and 70 percent stay with Zymeworks. The gap between what came in and the maximum that goes out is roughly $236.6 million — nearly three times the $80.4 million of stockholders' equity reported on June 30, 2026. How fast the math works depends on one number: royalty revenue, which was $1.8 million in the second quarter of 2026.

On April 8, 2026, FreeCast issued 137 warrants to 137 accredited investors, together covering 6,743,587 Class A shares at $4.25 each, expiring May 15, 2026. On May 8, 2026 the board cut the exercise price to $1.33 and pushed the expiry to May 22, 2026.

The outcome sits in the Form 8-K of May 28, 2026: two holders — Carl and Joyce Peterson — exercised and received 250,000 shares for $332,500 in total. The remaining warrants over 6,493,587 shares expired and the shares returned to authorized but unissued status. That is roughly 3.7 percent of what was originally offered. Two months later the same company raised roughly $23.7 million at $3.00 per share in a private placement — from institutional investors, not from the 137.

On April 20, 2026, FreeCast renewed the revolving convertible note held by Nextelligence, the chief executive's company, extending it to June 30, 2027. In the same move the conversion price changed: a fixed $8.00 per share became, per the prospectus, "the closing price of a share of Class A common stock on the Nasdaq Global Market on the most recent trading day prior to the date Nextelligence gives us written notice of conversion" — in other words the previous day's close.

The difference is fundamental. A fixed conversion price caps the number of new shares. A conversion price that follows the market creates more shares the lower the stock trades — and the lender picks the moment. Immediately after the renewal, Nextelligence converted $1,714,052 into 484,354 shares (at $3.51 and $4.00). As of July 15, 2026, $3,679,451 remained outstanding under the renewed note, against a $5 million facility at 12 percent interest and 18 percent default interest. Proceeds from the July 2, 2026 private placement may not contractually be used to repay it.

Between October 9 and November 21, 2025, Nextelligence, Inc. handed FreeCast, Inc. a total of $1,500,000. Nextelligence is majority owned by founder and chief executive William A. Mobley, Jr., who is also its CEO and sole director. Of that sum, $191,023 was, in the words of the prospectus, remitted "on behalf of Celebrity Cigars, Inc. and Test Drive Live Inc." in order to fully satisfy their outstanding receivables with FreeCast. Mobley is president of both of those companies as well, and sole director of Celebrity Cigars; his son Sean Mobley is part of its management team.

So one company owned by the chief executive settled the overdue bills of two other companies owned by the same chief executive at the publicly traded one. The remaining $1,308,977 was booked as a convertible note. For scale: those two customers together accounted for more than 35 percent of total revenue of $628,149 in the year ended June 30, 2025, and more than 52 percent the year before. The arrangements with both companies have been, per the prospectus, verbal since June 2023.

Anyone calculating cash per share at Atrium Therapeutics usually divides by 17,105,643 — the share count on the cover of the quarterly report filed August 13, 2026. Note 7 of that same report names a second number that is not included there: 1,590,677 shares promised to holders of former Avidity options and stock units as so-called Make Whole Awards, granted to compensate them for the spin-off. They had not been issued as of June 30, 2026 and must be settled, in the words of the report, "in no event after March 15, 2027."

That is 9.3 percent of additional shares, and they do not arise from a need for capital but from the mechanics of the separation: not a single dollar of new money comes in for them. Adding the 962,000 options and 431,000 stock units excluded as anti-dilutive as of June 30, 2026 takes the count to roughly 20.09 million shares — 17.4 percent more than today. Every per-share figure, including the frequently quoted cash per share, is therefore overstated on a horizon of months.

On August 14, 2026 three ownership filings on Atrium Therapeutics reached the SEC, all as of the June 30, 2026 record date — and two of them tell opposite stories. RA Capital Management, one of the best-known funds specializing in biotechnology, had reported 864,102 shares, or 5.6 percent, as of March 31, 2026. In the update as of June 30, 2026 the same line reads zero: the position was closed out entirely, in the first full quarter after the shares began trading.

On the same day Sessa Capital IM, L.P. of New York reported 1,500,000 shares, or 8.8 percent, for the first time — jointly with John Petry, filed under Rule 13d-1(b). That makes Sessa the largest single holder visible in ownership filings, ahead of T. Rowe Price with 1,472,984 shares (8.6 percent) and BlackRock with 1,135,757 shares (6.6 percent). With only 17,105,643 shares outstanding and average trading volume in the low six figures, 1.5 million shares is a position that cannot be moved unnoticed. The drug-development specialist is selling, a generalist fund is buying — the two clearly value the same company very differently.

In April 2026 the board rescinded the old repurchase authorization and established a new one: up to $300 million, effective May 14, 2026, with no expiration date (Form 10-K for fiscal 2026). Measured against a market value of roughly $1.35 billion (40,052,771 shares as of August 11, 2026 times the closing price of $33.71 on August 21, 2026), that is about 22 percent of the entire company.

At the same time the opening quarter of fiscal 2027 was the weakest in years: operating cash flow of $15.6 million against capital expenditures of $23.3 million, leaving free cash flow of negative $7.6 million (prior-year quarter: positive $17.8 million). Even so, $25.1 million went into buybacks and $9.7 million into dividends — $34.8 million combined, 62 percent more than a year earlier. Cash fell from $303.2 million to $267.3 million in three months. The program is affordable — the cash balance carries it — but right now it is being paid out of the stock of cash, not out of the operating business.

Joybird is La-Z-Boy's answer to online furniture retail — its own brand, sold digitally, with 16 small stores. What the project has cost so far is not in the narrative but in a footnote beneath the goodwill table of the Form 10-Q for the quarter ended July 25, 2026: "Includes $46.9 million and $20.6 million of accumulated impairment losses in Corporate and Other and the Wholesale segment, respectively." The $46.9 million is the running total of write-downs on Joybird goodwill. What remains is $35.5 million, and the Joybird trade name itself was fully amortized as of July 25, 2026.

The fiscal 2026 impairment alone came to $20.0 million — equal to 15 percent of that year's entire operating income of $129.2 million. The trend heading into the next test still points down: Joybird sales of $130.8 million in fiscal 2026 (down 10 percent), $26.5 million in the first quarter of fiscal 2027, with written sales down 17 percent. The company is also closing the Joybird plant in Tijuana and moving production to U.S. plants by the end of fiscal 2027. La-Z-Boy runs its scheduled goodwill impairment test in the fourth quarter of each fiscal year.

La-Z-Boy's pitch is simple: we build at home, so tariffs barely touch us. In fiscal 2026 roughly 90 percent of the upholstered units sold in North America were produced in the United States, and imported finished goods accounted for just 5 percent of consolidated sales (Form 10-K for fiscal 2026). A risk factor in the same filing flips the story around: "Conversely, if certain tariffs are eliminated or reduced, we may face additional competition from foreign manufacturers entering the United States market and from domestic retailers who rely on imported goods, putting pressure on our prices and margins."

This is no longer hypothetical. In February 2026, according to the annual report, the U.S. Supreme Court invalidated certain tariffs imposed under the International Emergency Economic Powers Act. La-Z-Boy is eligible for refunds but recorded none in its fiscal 2026 financial statements. In the first quarter of fiscal 2027 the favorable tariff impact — refunds plus pricing actions net of tariff costs — lifted the Wholesale segment gross margin by 240 basis points. On segment sales of $322.9 million that is roughly $7.8 million; without that tailwind the segment operating margin of 2.1 percent would arithmetically have been negative. In short: a one-time refund flatters today's margin, and the very tariff relief that produces it erodes the moat.

Kestrel Group's board has seven seats. Who fills them is set out in two agreements dated May 27, 2025 and described in the proxy statement (DEF 14A) filed April 24, 2026. Kestrel Intermediate Ledbetter Holdings (KILH), the founding family's holding company, may nominate two non-independent and two independent directors. AmTrust Financial Services may nominate one non-independent and two independent directors. Four plus three is all seven.

There is a second peculiarity that is rarely seen. The group's own subsidiary, Maiden Reinsurance, held 22.2 percent of its parent's outstanding shares at June 30, 2026. For accounting purposes those shares are treated as treasury stock and excluded from book value and earnings per share — but they still carry votes: the former 9.5 percent voting limitation was removed by shareholders on April 29, 2025. A quarter of the votes therefore sits with a subsidiary run by management. The free float stood at roughly 1.6 million shares on August 21, 2026 (fundamental data).

Kestrel Group had about 44 employees as of March 6, 2026. It granted its executives no equity awards for 2025 — and then made up for it twice over in 2026. On March 10, 2026 the compensation committee approved, according to a Form 8-K, restricted share awards of $650,000 each for fiscal 2026 plus $1,300,000 each as a catch-up for fiscal 2025 to Terry Ledbetter, Bradford Luke Ledbetter and Patrick Haveron. On May 13, 2026 a further $650,000 each followed in performance-based shares, equal to 61,588 shares per executive. In total, roughly $7.8 million of equity grants to three people within one year.

For scale: the entire Program Services segment produced fee income of $4.0 million in the first half of 2026, and the market value stood at roughly $60 million on August 21, 2026 (fundamental data). The 113,390 restricted shares granted to each executive in March plus the 61,588 performance shares add up to 524,934 shares — 6.7 percent of the 7,824,030 shares outstanding. The effect is already visible: diluted book value per share fell to $13.31 at June 30, 2026 while the undiluted figure was $14.57 — a gap of 8.6 percent, against 1.8 percent at the end of 2025.

On June 2, 2026 an arbitration between a subsidiary of Kestrel's Genesis Legacy Solutions (GLS) unit and a ceding company ended. The panel largely sided with Kestrel: the cedant had changed its reserving and claims administration practices after inception without the reinsurer's consent — an intentional and material breach of the reinsurance agreement. Kestrel recovered $5.25 million of the $10.8 million it had previously paid and was awarded $1.0 million in attorneys' fees.

The real finding sits in the small print of the Form 10-Q for the period ended June 30, 2026. The contract continues, and Kestrel remains exposed to the full limits of coverage. The filing states that the company may recognise additional losses up to and including the full limits of its exposure, less amounts presently reserved — quantified at a maximum of $28.2 million. That is roughly 25 percent of shareholders' equity of $114.0 million on the same date. The related balance sheet liability already rose from $4.0 million to $9.2 million in the first half of 2026. Kestrel says it is waiting for updated information from the cedant before it can size the impact: a case won, with the bill still open.

On March 11, 2026 Zvi Schreiber, the founder of Freightos and its chief executive until January 2026, filed a Schedule 13G — the form for holdings that explicitly do not aim to influence how a company is run. Fifteen weeks later, on June 29, 2026, the same holding was refiled on a Schedule 13D. In U.S. filing practice that is not a change of paperwork but a declaration of intent: 13D is the form for shareholders who want something to change.

The reported position is 3,131,931 shares, or 6.1 percent of the company. In Item 4 Schreiber writes that the strategy pursued since the first quarter of 2026 is impairing shareholder value, that he wants the company returned to a platform-first growth strategy, and that he wants to change the composition and leadership of the board. He states that he intends to put proposals to the annual general meeting, and if necessary to solicit proxies or litigate — and that he has already held discussions with third parties "including potential strategic or financial acquirers".

Against a market capitalization of roughly $70 million (data as of August 21, 2026), a 6.1 percent block in the founder's hands is the single largest source of unrest around this stock — and the only documented indication that anyone has discussed a sale of the company.

In the list of credit facility covenants in the 2025 annual report (Form 20-F), sitting directly after two entirely ordinary financial ratios, is a third clause you rarely see spelled out. Alongside a minimum net worth of $150 million and an interest cover of at least 2.0, the secured facilities require that at least 30 or 35 percent of the voting and ownership rights remain with the Hajioannou family; under one facility, chief executive Polys Hajioannou must personally hold at least 20 percent (the “Control Covenant”). Selling the family stake below those thresholds would therefore not merely be an ownership matter — it would be an event of default on facilities that stood at $397.7 million of secured debt as of June 30, 2026.

Two things are moving toward that line. First, on May 18, 2026 Vorini Holdings Inc. reported in amendment No. 7 to its Schedule 13D a restructuring of control within the family: Nicolaos Hadjioannou no longer holds voting or dispositive power over Vorini's 19,426,015 shares, and Vorini is now controlled solely by Polys Hajioannou. In total, the six family entities held 48,381,427 shares, or 47.5 percent, as of May 15, 2026. Second, a buyback program for up to 10,000,000 shares has been running since December 2025; 515,469 shares had been repurchased and cancelled as of July 24, 2026. If the program were completed in full, the family stake would rise on arithmetic alone to roughly 52 percent — without the family buying a single share.

Note 4 of the quarterly report to June 30, 2026 puts the total consideration for the three businesses acquired on November 14, 2025 at $22.6 million — $5.4 million in cash, $15.9 million in 2,672,013 shares and $1.4 million of contingent consideration. The sentence that follows is the interesting one: the arrangement provides for up to $25 million in earnout payments, "payable in cash or stock", if the acquired businesses hit certain revenue targets over five years. The possible top-up is larger than the entire price already paid.

Almost none of that appears on the balance sheet. Note 2 carries the contingent consideration at $1.553 million as of June 30, 2026 (December 31, 2025: $1.382 million) — roughly six percent of the maximum. That is not sleight of hand but ordinary probability-weighted fair value: it says management currently views those revenue targets as largely out of reach. Which is exactly why this one line is a leading indicator. If it climbs materially in coming quarterly reports, the acquired businesses are performing better than the purchase price assumed — and it gets expensive at the same time, because each step-up runs through earnings and the company may settle in its own stock. The prospectus of January 21, 2026 has already registered up to 5,693,433 shares for resale on that account. For scale: total stockholders' equity at the balance sheet date was $70.843 million.

Note 17 of the interim statements to June 30, 2026 describes something rarely seen in a quarterly report: right-of-use assets of $5,820,618 were added during the quarter — and fully depreciated within that same quarter. Both the right-of-use asset and the lease liability stand at nil at the balance sheet date. The liability was not settled in cash: $5,916,434 was settled using digital currency, that is, with mined bitcoin.

That explains two things at once. First, the 79.2 percent jump in direct costs, which the company itself traces to this single line in section 7(b) of its MD&A — without it, direct costs for the quarter would have been $6,591,179, below the $6,926,572 of the year-earlier quarter. Second, the fall in the bitcoin balance: a substantial part of the 286.04 coins disposed of over the nine months went into this settlement. For scale, the $5.9 million equals roughly 89 percent of total equity of $6,615,513 at the balance sheet date. Whether Bitzero rents that capacity again in the fourth quarter decides whether the gross result turns positive.

Note 23 of the interim statements to June 30, 2026 lists two live disputes. In the first, Bitzero is in legal proceedings with its former Chief Executive Officer over employment matters and equity instruments; he has filed a counterclaim for damages. Outcome and magnitude cannot presently be determined and no provision has been recorded.

The same former chief executive is also a creditor: note 18 discloses an unsecured convertible loan with a carrying amount including accrued interest of $1,752,590 (contractual principal $1,000,000). The loan and the related conversion right are part of the same dispute. Section 18 of the MD&A states that the number of voting common shares potentially issuable under that conversion right cannot presently be determined — the outstanding securities table simply leaves that cell empty. Against equity of $6,615,513, both a quarter of book value and an unquantified amount of dilution therefore hang on a court case.