Have you ever watched a company race toward a public listing and wondered who will actually call the shots once the ticker starts trading? That question is sitting on a lot of desks right now. Anthropic, the company behind Claude, is asking shareholders to approve a plan that would give chief executive Dario Amodei and six other co-founders a combined 50.1% of voting power after a planned IPO. The economic slice they hold would stay far smaller. In my experience, that gap between cash ownership and voting muscle is where public investors either get comfortable or start asking harder questions.
What The Founders Are Asking Shareholders To Approve
The company is seeking approval for a special class of shares before it goes public. Those shares would let seven co-founders keep majority voting power on most corporate matters, so long as at least three of them still hold a minimum number of company shares. Each founder currently owns about 2% of the firm. The extra votes would not enlarge their economic stake. They would simply weigh more when ballots are counted.
That is the heart of the story. Control without a matching slice of the residual cash. I’ve found that readers often treat “majority control” as if it always means “majority ownership.” It does not. Dual-class and special-vote structures exist precisely to break that link. Whether you like that design or not, it is already common among large technology listings. This proposal sits in that tradition, with a few twists that are very much Anthropic’s own.
Shareholder approval is still required. Nothing here is locked. The plan also leaves room for the arrangement to fade if founders sell down below the still-undisclosed minimum. That sunset-by-ownership idea matters. It is not an eternal throne. It is a conditional lock on the steering wheel.
Why Fifty Point One Percent Is The Magic Number
Fifty percent plus a sliver is enough to win most ordinary shareholder votes. You do not need seventy. You do not need ninety. You need just over half. Once you clear that line, you can usually decide routine corporate questions without assembling a coalition of outside funds. That is why 50.1% shows up again and again in control documents. It is tidy. It is decisive. It is also easy for lawyers to draft around.
Perhaps the most interesting aspect is how little equity that threshold can rest on. Seven people at roughly two percent each would own about fourteen percent of the economics if those figures still hold at listing. Fourteen percent of the cash flow claim, more than half of the votes. Public buyers would supply most of the capital and still sit in the minority on many matters. That trade-off is not new. It is simply being restated at a scale that makes people sit up.
Shares with extra votes can let founders steer a company without owning most of its equity, leaving public shareholders with less influence over corporate decisions.
U.S. listing guidance already warns investors to read the prospectus for share classes and voting rights. That is not legal theater. It is the only place the mechanics will be spelled out with the precision markets need. Until that document is public, every number in this conversation is still a sketch.
A Structure That Echoes Other Founder-Led Listings
People close to the planning have compared the idea to founder-control models used by other well-known public companies. The resemblance is real enough. Super-voting stock. A trigger tied to continued ownership. A small group that can outvote a much larger crowd of ordinary holders. If you have followed technology IPOs over the last decade, this will not sound exotic.
What is more distinctive here is the second layer. Anthropic is a public benefit corporation. It also has a Long-Term Benefit Trust that elects directors. Founder votes and trust votes would not sit in the same pocket. That split is easy to miss if you only read the headline about 50.1 percent. I keep coming back to it because governance fights rarely happen on one axis. They happen when two axes collide.
How Board Elections Would Stay Separate From Founder Votes
Under the reported plan, the Long-Term Benefit Trust would still choose a majority of directors. Founder-elected seats would rise from two to three. The board has seven seats, one of them vacant. Do the arithmetic. Majority control of the board would not automatically move to the founders even if they win most shareholder votes on other topics.
That is unusual enough to linger on. In a classic dual-class company, the same people who dominate the annual meeting also dominate the slate. Here, the trust keeps the bigger lever on who sits in the room. Founders get a louder voice among shareholders and one extra board chair. Employees would receive another special class that could break ties on some decisions. Their role would not steal the trust’s power to elect directors. It would sit off to the side, useful in a deadlock, limited in scope.
I’ve found that investors often flatten all of this into “founders in charge.” That shortcut is sloppy. Charge of what? The annual vote on ordinary business? The composition of the board? The mission language in the charter? Those are different rooms. Different keys.
- Founders would hold 50.1% of votes on most shareholder matters if the minimum-holding test is met.
- The trust would keep the power to elect most directors.
- Founder board seats would increase from two to three on a seven-seat board.
- A separate employee share class could break ties on defined issues.
- Public ordinary shares would still carry economic rights and weaker votes.
What The Long-Term Benefit Trust Is Designed To Do
Anthropic describes itself as a public benefit corporation. The trust is presented as an independent body whose members have no financial stake in the company. It holds a separate class of stock that lets it elect and remove directors, with that authority designed to grow toward a majority of board seats. When the trust was set up, the company said the board would still oversee major decisions. Outside voices would help pick the people in those seats.
On paper, that is a check. In practice, checks only work if the people holding them use them. I am not going to pretend a trust is a magic wand. It is a legal device. Its force depends on the charter, the people appointed to it, and how conflicts get resolved when founders, employees, and public holders disagree. The details of those intersections will matter more than any slogan about “mission.”
Still, the design is not empty. It tries to keep director selection from collapsing entirely into founder voting power. If you care about long-horizon safety work in frontier models, that separation is the feature, not a footnote. If you care about tidy shareholder democracy, it is a complication. Both readings can be true at once.
Public Benefit Status And Why It Changes The Conversation
A public benefit corporation can pursue a stated public purpose alongside profit. Directors get a wider mandate than “maximize near-term equity value and nothing else.” That does not mean they ignore returns. It means they can weigh other aims without automatically breaching duty. For a lab building general-purpose systems, that legal wrapper is part of the brand.
Combine that wrapper with super-votes and a trust, and you get a company that is hard to push around from the outside. Some investors will call that stability. Others will call it insulation. I tend to think both camps are describing the same machine from different ends of the hallway.
The reported proposal would put two kinds of control in different hands. Founders would hold just over half the votes on most shareholder matters, while the trust would select most directors.
What This Means For Ordinary IPO Buyers
If you buy stock in a listing like this, you are usually buying economics first and influence second. Dividends, if they ever appear. Price appreciation. Information rights. A vote that may not move the needle. That bargain is fine if you go in with eyes open. It is a problem if you assume one share equals one voice and then discover the fine print after allocation.
For U.S. investors, the prospectus is the place to check share classes. Look for conversion rights. Look for sunset clauses. Look for the exact minimum holding that keeps founder super-votes alive. Look for how employee tie-break shares work. Look for whether the trust can expand, shrink, or be rewritten after listing. Those pages are dry. They are also the whole game.
Would I personally treat a thin-vote share as “worse” than a full-vote share? It depends on the price. Control discounts exist for a reason. If the valuation already prices in limited influence, the math can still work. If the valuation assumes you are buying a normal governance package, you are paying for a product you are not getting. That is not ideology. That is shopping.
| Group | Economic Stake | Voting Role |
| Seven co-founders | About 2% each today | Proposed 50.1% on most matters |
| Long-Term Benefit Trust | No personal financial stake for members | Elects most directors |
| Employees with special shares | Varies | Possible tie-break on defined issues |
| Public ordinary holders | Majority of float after listing | Minority on many votes |
Valuation Talk Is Still A Moving Target
The voting proposal arrives while the size and timing of a potential offering keep shifting in market chatter. Recent discussion has pointed to a possible November window, a raise that some investors have talked about in the neighborhood of $100 billion, and a headline value near $2 trillion. Those terms were described as preliminary. Treat them that way.
A more anchored figure comes from the last announced private round. In May the company said it raised $65 billion in Series H financing at a $965 billion post-money valuation. It also said annualized revenue had crossed $47 billion earlier that month. Secondary-market estimates later clustered around $1.5 trillion. Private prints do not set the IPO price. The offering price will depend on how many shares are sold and what buyers will actually pay that week.
Confidential filing for a U.S. listing has been reported earlier. Marketing calendars have slipped before and they will slip again if markets wobble. Late September for a public prospectus and mid-October for a roadshow have been floated in prior coverage, always with the usual caveat that dates move. Anyone treating those dates as carved in stone has not watched many offerings.
Pre-IPO Contracts Are Not The Same Thing As Shares
Interest in the listing has spilled into crypto-linked trading venues. Some platforms offer perpetual futures or similar contracts tied to an implied private-market valuation. Eligible traders get price exposure. They do not get Anthropic shares. They do not get dividends. They do not get votes. U.S. customers are often shut out of those products. Other venues have rolled out related contracts for certain European clients, in some cases with leverage up to ten times.
Those prices can drift away from both the last funding round and any eventual IPO print. Leverage makes that drift louder. I would not confuse a contract that tracks a number with a security that confers membership in the cap table. One is a bet. The other is ownership, even if the ownership comes with a weak vote.
Why mention this at all in a governance piece? Because a lot of retail attention now arrives through these side doors. People talk as if they are “in” a company when they are only in a derivative. When the prospectus finally lands, that distinction will matter. Votes live on the share register. They do not live in a perpetual.
The Investor Case For Founder Super-Votes
There is a serious argument in favor of this design. Frontier model companies make decisions that are hard to reverse. Training runs are expensive. Safety policies are contentious. Talent is mobile. A founder group that can hold a line through a noisy public market may avoid the quarterly whiplash that turns research labs into slide-deck factories. If you believe the product is a multi-decade platform, you may want fewer hands on the wheel, not more.
Super-votes can also keep strategic buyers and activists from forcing a sale the board does not want. In a hot sector, that is not a hypothetical. Control premia exist because control is valuable. Founders are, in effect, asking public markets to finance growth while leaving that premium with the original team and the trust.
I’ve sat through enough listing conversations to know that many long-only funds will accept this if the growth story is strong enough. They grumble. They still allocate. The complaint shows up later, usually after a bad quarter, when people suddenly remember they cannot easily replace management through a vote.
The Investor Case Against It
The case against is just as plain. Capital should come with a voice. If the public float supplies most of the equity value, the public float should not be a decorative gallery. Entrenchment can protect a good mission. It can also protect a stale one. There is no clause that guarantees wisdom. There is only a clause that makes removal harder.
Related-party deals, compensation, and acquisition strategy all get stickier when the people proposing them also dominate the vote. Even a well-run trust does not erase that tension. It relocates it. And if the minimum-holding test is modest, founders could keep control after selling a large part of their economic exposure. Alignment thins out. Incentives get weird. That is not a cartoon villain story. It is a known failure mode of dual-class stock.
- Read the exact voting multiple on each class.
- Find the ownership floor that keeps founder control alive.
- Map which decisions fall to shareholders versus the board versus the trust.
- Check how employee tie-break shares are issued and retired.
- Price the stock as a limited-influence claim, not as a town-hall ticket.
How Dual-Class Listings Usually Age
Some companies keep super-votes for decades. Some let them lapse when founders leave or fall below a threshold. Some convert on a time clock. Markets have seen all three. Time-based sunsets are cleaner for index inclusion debates. Ownership-based sunsets are more personal. They last as long as the original group stays invested enough to care.
Anthropic’s version, as described, leans on the ownership test and a three-founder minimum. That is a group lock rather than a single-person lock. If two founders exit and five remain above the line, control can persist. If the group shrinks below three, the special votes weaken or vanish. The missing number is the per-person floor. Until that figure is public, you cannot model how easy it is to keep the club intact after diversification sales.
Index providers and governance raters will have their own views. Some funds will not buy dual-class names at all. Others run dedicated sleeves for them. None of that settles the policy debate. It just tells you the buyer universe is already split.
Employees As A Quiet Third Force
The employee tie-break class is easy to shrug off. Don’t. In a close fight over a defined corporate action, a block that exists only to break ties can punch above its weight. It will not elect the board. It will not rewrite the trust. It can still decide a contest that founders and ordinary holders split down the middle.
Who receives those shares, how they vote, and whether they move as a bloc will shape real outcomes. Employee interests are not identical to founder interests. They are not identical to public-fund interests either. Compensation, culture, and product caution can pull them in different directions. A tie-break class is a small valve. Small valves change pressure when the system is tight.
What The Prospectus Will Need To Make Clear
A public prospectus would give buyers firmer detail on the share structure, financials, and offering terms. That is the document that turns rumor into a contract with the market. I would want, at minimum, a clean map of every class, a worked example of a contested vote, and a plain-language description of how the trust interacts with founder super-votes when those two powers point different ways.
Risk factors will almost certainly flag limited public influence. They should. Buyers who skip that section are volunteering for surprise. Underwriters will also have to explain why the structure is consistent with the company’s public-benefit purpose rather than a simple entrenchment device. The words will be polished. Read them twice anyway.
Control map to watch: Shareholder votes - founders at 50.1% if conditions hold Board majority - trust-selected seats Founder seats - rising to three of seven Tie-break valve - employee special class Economics - mostly with public and institutional float
A Few Practical Notes For People Watching The Clock
First, approval by current holders is a separate event from the IPO itself. The company can win the vote and still delay the listing. It can also reshape the proposal if large existing investors push back. Second, valuation headlines will swing harder than governance headlines. That does not make governance less important. It just means the market’s attention span is uneven.
Third, secondary prints and derivative contracts will keep inventing a daily “price” even when no new primary share is sold. Those numbers are useful as mood rings. They are poor substitutes for a book-built offering. Fourth, if you already work at the company, your personal tax, lockup, and vote profile may look nothing like the public story. Talk to counsel before you treat a news item as a personal plan.
In my view, the cleanest way to follow this is to ignore the loudest valuation rumor of the week and keep a short checklist: Did current shareholders approve the class? What is the minimum holding? How many founder seats sit on the board the day after listing? Does the trust still elect a majority? Until those four answers are in a filed document, everything else is atmosphere.
Why This Story Travels Beyond One Company
Every large model lab that lists will face a version of this problem. The work is capital hungry. The downside scenarios are unusual. The founders want continuity. Public markets want a claim they can understand. Dual-class stock is the standard compromise of the last fifteen years. Trusts and benefit-corporation wrappers are the newer layer. Anthropic is not inventing founder control. It is stacking extra institutions on top of it.
Regulators have lived with dual-class listings for a long time. They require disclosure more than they ban the tool. That is unlikely to change on the timetable of one offering. What can change is buyer appetite. If the first few AI listings with heavy control discounts trade poorly, the next ones will pay for it. If they trade well, the structure becomes a template. Markets teach by marking prices, not by writing essays.
I keep thinking about the quiet part. Seven people. A trust with no personal economic stake. Employees with a tie-break chip. A public float that funds the dream and sits in the cheap seats for many votes. That is a lot of architecture for a company that still has to ship product, manage costs, and survive a cycle. Governance is not the business. It is the frame around the business. Frames matter when the picture gets heavy.
The Bottom Line Before The Filing Lands
Anthropic is asking current shareholders to bless a special class that would give seven co-founders just over half the votes on most matters after a public listing, provided at least three of them keep a minimum stake. Their economic ownership would stay far below half. The trust would still pick most directors. Founder board seats would rise by one. Employees could break certain ties. The IPO price, date, and size remain unsettled even as private-market figures stretch from the last $965 billion round toward much larger rumored prints.
None of that is a verdict. It is a design. Designs have trade-offs. If you want continuity and a mission wrapper, this package is coherent. If you want ordinary-share democracy, it is a polite no. The honest move is to decide which product you are buying before the roadshow music starts. Prospectuses are long for a reason. This is one of those reasons.
And if you only remember one number from this whole debate, make it 50.1. Not because it is poetic. Because it is the line between influence and audience. On one side of that line you help steer. On the other you watch the car and hope the drivers like the same destination you do.
