How the Aoki Family Won a Poison Pill Vote It Had Already Paid For

Two numbers from the same company's audited accounts explain the control fight at Kusuri no Aoki Holdings. The grant-date fair value of one share under the fifth series of its share acquisition rights, as the company itself disclosed it, was ¥691. The amount the holders paid for that right was ¥5.^K12K11

The poison pill approved on February 17, 2026 is administration, and so are the buyback and the move off the Tokyo Stock Exchange's Prime Market that surround it. By the time proxy advisers were writing reports and Bloomberg was calling Ishikawa Prefecture, the ownership question had been settled for eighteen months by a board resolution taken in January 2020 and a set of shares issued in August 2024.

The company disclosed the vote result as an approval rate of 55.5%, with a footnote saying the figure was approximate as at that date and that a more detailed rate would appear in its extraordinary report.^K2 A margin of five and a half points is not usually the sort of thing a winner rounds.

The useful reading of this fight is not whether the family deserved to win. It is where the winning actually happened, because the place where it happened is available to almost any founder with a cooperative board and a quiet January.

Kusuri no Aoki Holdings runs 1,144 stores from Hakusan in Ishikawa Prefecture, traces itself to a drug seller founded in 1869, and generated ¥566.8 billion of revenue in the year to May 2026.^K14K1 Its president, Hironori Aoki, is the sixth generation of the family. Bloomberg put the family's holding at roughly 36%.^B1

The sequence that produced that holding runs in this order. In January 2020 the board resolved to issue 35,000 paid share acquisition rights to one director of the company and one director of a subsidiary.^K11 In August 2024 all of them were exercised, adding 10,500,000 shares.^K12 Between October 2024 and December 2025 the company ran successive buyback programs and cancelled 9,480,700 shares.^K13K14 In December 2025 it applied to leave the Prime Market.^K9 In January 2026 its twenty-three-year capital and business alliance with Aeon ended, the Aeon-nominated outside director resigned, Aeon said it would stop seconding a director, and the defense measure was announced.^K7K8A2K1 In February 2026 shareholders approved it.

Read forward, this is a company reacting to a threat. Read backward, it is a company finishing a capital structure and then buying insurance on it.

The step that decided the outcome attracted the least scrutiny. The defense measure got an extraordinary general meeting, a proxy fight, and voting recommendations against it from both Institutional Shareholder Services and Glass Lewis.^B1 The option grant that produced 10,500,000 shares is recorded in the filings as a board resolution dated January 9, 2020.^K11 Two decisions, one register, and the ballots went to the smaller of them.

The fifth series of share acquisition rights is documented in the company's own filings in enough detail to reconstruct the whole thing.

The board resolved on January 9, 2020 and granted on January 28. Recipients: one director of the company and one director of a subsidiary. Volume: 35,000 rights over 3,500,000 shares before the November 2023 three-for-one split, restated by the company as 10,500,000 shares. Exercise price: ¥6,830 before the split, ¥2,277 after. The report records an issue price of ¥6,845 per share against that ¥6,830 exercise price, so the holders paid ¥15 per pre-split share for the option itself, which is ¥52.5 million across 3,500,000 shares.^K11K12

Then the number that decides the argument. The same report discloses the grant-date fair unit value of the fifth series as ¥2,073, calculated by Black-Scholes on 33.76% volatility and a 7.25-year expected term, and the FY2025 note restates it after the split as ¥691 per share.^K11K12 The holders paid ¥15 for something the company's own appraisal put at ¥2,073. Across the whole grant, that is ¥52.5 million paid against ¥7.26 billion of disclosed value.

Paid stock options with performance and price conditions are an ordinary Japanese instrument, and their pricing logic is legitimate. Harsh conditions reduce fair value. Here the conditions were genuinely harsh on paper. The rights could only be exercised if ordinary profit exceeded ¥22.0 billion in one of the six years to May 2029, they died if ordinary profit fell below ¥11.0 billion first, and they died if the forty-two-day average closing price dropped below 70% of the exercise price at any point before May 20, 2024.^K11K12

One clause deserves a slow read. The profit test was run on ordinary profit before deducting goodwill amortisation and before deducting the stock compensation expense of these same options.^K11 The company recognised ¥6,810 million of stock compensation expense in the year to May 2024 and ¥392 million the following year.^K12 Reported ordinary profit for the year to May 2024 was ¥20,101 million, which is ¥1.9 billion below the ¥22.0 billion hurdle.^K14 The exam was graded on a version of the accounts with the cost of the prize erased from the paper.

At exercise the average market price was ¥3,122.^K12 Two directors acquired 10,500,000 shares at ¥2,277 in a company where the shares were trading above ¥3,100, having paid ¥52.5 million five years earlier for the right to do it.

Oasis Management, which manages funds holding Kusuri no Aoki shares, has run a shareholder derivative suit since July 2024 seeking roughly ¥7.2 billion from Hironori Aoki and Takanori Aoki, alleging the options were sold at a discount of more than 99% to the fair value the company itself disclosed.^O1 Those are allegations in pending litigation, and the company disputes Oasis's characterisation of its capital policy.^K6 The valuation, the ¥52.5 million, and the exercise price are not allegations. They sit in an audited note that any shareholder could have opened in 2020, four years before a single share was issued, in a document nobody reads because it is the part of the annual report that comes after the pictures of the stores.

The option exercise created 10,500,000 new shares, which diluted everyone else. What happened next removed the dilution from the share count without removing it from the register.

Under the program resolved on July 3, 2025, the company bought 5,609,200 of its own shares for ¥21,999,623,500, an average of about ¥3,922 a share, and completed the frame in December.^K13 On November 20, 2025 it cancelled 9,480,700 shares.^K14 Issued shares fell from 105,097,680 to 95,616,980, and shares outstanding excluding treasury fell from 101,102,991 to 94,963,067.^K12K14K13

Now hold the family's holdings still and watch the percentage move. The seven limited liability companies through which the family appears in the top-shareholder tables held 31,738 thousand shares at May 20, 2025 and 31,734 thousand a year later, a difference inside the rounding.^K12K14 As a share of stock outstanding excluding treasury, that block went from 31.39% to 33.42%. Two points of a listed company, acquired by not selling, paid for with the company's cash.

This is the least discussed control mechanism on any cap table. A buyback is compulsory for the buyer and optional for the seller, which makes it a continuous transfer of ownership from whoever will sell to whoever will not. The holder least likely to sell is the founder. Describe it as capital efficiency and it clears the board pack. Describe it as a control transfer financed by the company and it becomes a different agenda item, which is why nobody does.

Notice what is absent from this structure. There are no dual-class shares here and no Control Wedge between votes and economics. Every vote the family casts sits on top of a share it genuinely owns. That is exactly what makes the arrangement durable: a wedge invites a sunset clause, an index exclusion, and a decade of investor letters, while ordinary common stock acquired cheaply invites nothing at all.

The two transactions sit awkwardly beside each other. The company issued 10,500,000 shares to two directors at ¥2,277 and then spent ¥22 billion buying 5,609,200 shares back from the market at an average of ¥3,922. Both were disclosed as steps toward improving shareholder value, and in the narrow accounting sense both were.

Concentration has a listing consequence. The Tokyo Stock Exchange requires a tradable-share ratio of at least 35% for continued listing on the Prime Market. On the Standard Market the requirement is 25%.^J1

On December 25, 2025, the same board meeting that set the record date for the extraordinary shareholders' meeting also resolved to apply for a move from Prime to Standard and a new listing on the Nagoya Stock Exchange Main Market. The stated purpose was to prioritise medium- and long-term growth and management efficiency under a more appropriate market segment.^K9

Aeon read that differently and said so on January 15, 2026: a large volume of stock options had been exercised in 2024, the founding family's holding had risen from about 27% to 40%, the tradable-share ratio had fallen accordingly, and moving to Standard would let the company stay listed while the family kept its ratio.^A2 The company's board answered on February 16, the day before the vote, saying the option exercise was not the direct cause of the market segment change and that attributing the float decline to one shareholder's conduct was misleading.^K6

On June 12, 2026, the company announced the approval and gave its own reason. The purpose of the application, it said, was that meeting the tradable-share ratio standard required of a Prime-listed company had become difficult.^K10

Four months separate the rebuttal from the confirmation. The February document was written for people who had not yet voted, and the June document was written for people who already had. Both are accurate. Only one of them was load-bearing.

The transferable point is not that the company misled anyone. It is that a listing standard looks like a constraint and behaves like a preference. If your ownership concentrates past a threshold, you can restore the float or you can relocate to a venue with a lower threshold, and only one of those options is under your sole control.

The defense itself is conventional in shape and specific in aim.^K1

A "large-scale share acquisition" means any purchase intended to take a specific shareholder group to 20% or more of voting rights, or that results in that outcome. An acquirer must file a letter of intent, receive an information list within ten business days, and answer it, with a final deadline of no more than sixty days from the letter. Then the board takes sixty days to evaluate, or ninety where the consideration includes non-cash items, is payable in a currency other than yen, or the acquirer plans major changes to the group's structure. If the acquirer does not comply, the board may allot share acquisition rights carrying exercise conditions that exclude them, which is the dilution.

The clause that does the work is the one covering holders who are already there. Where a shareholder group is at 20% or more when the policy is adopted, or reaches it in combination with another shareholder, that group is treated as a large-scale acquirer, and the rulebook applies to any further purchase, including, in the filing's own words, the acquisition of a single additional share.^K1

At May 20, 2026, Aeon held 10.23% and Tsuruha, which Aeon controls, held 5.11%. Oasis Fundamental Value II (Hong Kong) held 14.08%.^K14 Neither group is over 20% on its own. Together they are at roughly 29%, and the policy's test for whether two holders are acting jointly runs on shareholding relationships, business alliances, board interlocks, funding, lending, the pattern of buying, and how they have voted.^K1 A defense drafted against hypothetical future raiders has an unusually precise fit around the two shareholders who have publicly disagreed with the board.

The safeguard offered against arbitrary use is an independent committee of at least three members, independent of executive management, of whom at least two must be outside directors.^K1 Read that independence test carefully: it is independence from the people running the company, not independence from the people who own it, and in this situation those are close to the same people.

The committee's composition then changed the day before the vote. The candidate slated to fill the third seat, Yoshiki Nishii, withdrew for health reasons, the board withdrew the resolution to elect him, and on February 16 it named Shinichi Morioka in his place.^K4 Morioka is a lawyer and was at the time the company's own substitute corporate auditor, a role the board cited as evidence he had already been vetted.^K4 Oasis had separately argued that the committee excluded the outside director who is a practising lawyer, Naoki Yanagida, and the company called that speculation.^K6K3

Nobody chose to be unwell, and the withdrawal was nobody's fault. What the episode exposes is the design. A body whose whole function is to constrain the board was reconstituted by that board, twenty-four hours before shareholders voted on whether to give it a dilution trigger, and there was no procedure that required anyone to pause.

The company also made a point worth conceding. It noted that the takeover defense Aeon's own shareholders approved in May 2024 can be triggered by Aeon's board alone, and that Oasis, which voted at that meeting, raised no objection there.^K6 That is a fair hit. Structural criticism of takeover defenses in Japan is applied unevenly, and being right about Kusuri no Aoki does not make anyone consistent.

Here is the arithmetic that the reported percentage hides, with the inputs stated so you can rerun it on your own numbers.

Approval was reported at approximately 55.5% of votes cast.^K2 Assume every family share voted in favour, which is safe at a meeting the family called. Aeon put the family at 40% of the company; Bloomberg's calculation put it at about 36%.^A2B1

Take the higher figure and assume every share on the register voted. The family supplies 40 of the 55.5 points in favour, leaving 15.5 points of support from a non-family electorate of 60 points, which is about 26% support outside the family. Take the lower figure on the same assumption and outside support rises to about 30%. Now relax the turnout assumption, because turnout is never 100%. At 85% of the register voting, with the family voting in full, the family is 47% of the votes actually cast, and outside support falls to roughly 16%.

On any plausible combination of those inputs, somewhere between about a sixth and a third of the non-family votes cast supported the measure. The proposal that "passed with 55.5% support" lost the room by a wide margin among everyone whose surname is not on the stores.

That is the part worth carrying into your own governance work. An approval rate that includes the proponent's own block is a fact about the register, not a fact about consent. It is the same arithmetic problem as a controlling-shareholder buyout where the majority-of-minority condition is measured against a padded denominator, and it produces the same false comfort. Japan's framework does not require a majority-of-minority condition on defensive measures. A family at 40% asking for approval needs 15.5 further points from a 60-point electorate, and can lose three quarters of that electorate and still be reported as having won.

The market read it that way too. Bloomberg reported that the shares fell as much as 9% after the result, the biggest intraday drop since December 2022.^B2

If you are a founder considering a shareholder-approved defense, or a director being asked to recommend one, work through these before anyone drafts a circular.

• Recompute the approval rate net of the proponent. Remove the founder block from both numerator and denominator and state that number in the circular. If it is below 50%, you know what you have: a mandate from the registrar rather than from the shareholders.
• Offer a majority-of-minority condition voluntarily. No Japanese rule requires it. Offering it is the single cheapest way to convert a contested defense into a credible one, and refusing to offer it tells everyone what the count is for.
• Fix the committee before the vote, and write its independence test against the actual conflict. Independence from executive management is the wrong test when the controlling shareholder is the executive management, so define independence against the person who benefits. Then say in advance who fills a vacancy, on what timetable, and whether a change inside a stated window postpones the meeting. A committee the board can reconstitute the day before a vote is an advisory body with a good title.
• Price the option grant, not the pill. If your board is being asked to approve performance options for insiders, model the fully diluted register at exercise, not the accounting expense. Then ask whether the performance condition is measured before or after the cost of the award itself.
• Model the buyback as a control transfer. For every repurchase authority, calculate the percentage move for each holder who will not sell. Put that column in the board pack next to earnings per share.
• Check your listing headroom before you concentrate. Work out what the free-float ratio becomes after every planned grant, exercise, and repurchase, and find out what your exchange requires. Discovering the answer afterwards leaves relocation as the only remaining fix.

This describes mechanics rather than advice on any specific transaction. Takeover defenses, directors' duties, and disclosure obligations differ by jurisdiction and by a company's own articles. Japan's framework here runs through the Companies Act, the Financial Instruments and Exchange Act, METI's 2023 guidelines on corporate acquisitions, and the exchange's own rules, and it does not work like Delaware. Test any real situation with qualified counsel.

It makes it harder to challenge, which is not the same thing. A vote converts a board decision into a shareholder decision and gives directors a much stronger record if the measure is ever litigated. What it does not do is tell you whether the people the measure protects against were outvoted or outnumbered. The diagnostic is one subtraction: take the proponent's block out of the count and look at what remains. A defense that survives that subtraction has genuine consent behind it. A defense that only clears with the proponent's own shares included has a procedure behind it.

The individual instruments travel. Performance options to insiders, buybacks that lift a static block, and a move to a less demanding listing venue all exist in most markets. What varies is where the friction sits. A US company running this sequence would meet Section 16 reporting, Schedule 13D disclosure by the shareholders it annoyed, exchange rules on equity compensation, and a plaintiffs' bar with a well-worn path to Chancery. A UK premium-listed company would hit related-party rules and free-float requirements sooner. So the question to ask about your own jurisdiction is not whether it permits the sequence. It is which step it actually polices, because founders tend to find that out only after they have tried both.

Kusuri no Aoki's twenty-eighth annual general meeting is on August 19, 2026 at a hotel in Kanazawa, with the voting deadline at 5:00 p.m. on August 18.^K16 Eleven directors are up for election. Takanori Aoki, the vice-president and the second recipient of the 2020 options, resigned from the boards of the holding company and its operating subsidiary on March 31, 2026 for personal reasons, six weeks after the meeting that secured the defense.^K17 He is not standing. Naoki Yanagida, the outside director who is a lawyer, is retiring. Two of the three new outside director candidates are Takeshi Yoshida, a former director of Hokuriku Bank, the group's main bank, whom the company notes left that board more than three years ago and therefore satisfies the exchange's independence criteria, and Shinichi Morioka, the substitute corporate auditor who joined the pill's independent committee the day before the February vote.^K15

Whatever that meeting decides, it will not decide the ownership question. That closed on January 9, 2020, when a board resolved to sell one of its own directors and one director of a subsidiary a claim on what became 10,500,000 shares for ¥52.5 million.^K11 If you want to know where control in your company will actually be settled, do not look at the agenda item marked "takeover defense." Look at the one marked "share acquisition rights," and read the note on fair value before you raise your hand.