There is a passage in SpaceX’s S-1 that, once you read it, makes you realize this might be the most candid filing of any large IPO in the past twenty years. It does not hide anything. It simply writes into the registration statement something that public companies normally do not say out loud. It sits in the Capital Stock section, and it reads:
Mr. Musk will only be subject to removal from the board and from his Chief Executive Officer and Chairman of the board leadership positions with the approval of the holders of at least a majority of the voting power of the outstanding shares of our Class B common stock, voting separately as a class.
Translated: whether Elon Musk can be removed is decided by a vote of Class B shareholders. And the voting power of Class B is in Musk’s hands.
That fact alone has already been covered by many media outlets. But the full mechanism in the S-1 is richer than the headline “Musk controls the company.” It is not a simple founder-retains-control arrangement. It is a carefully constructed closed loop. It specifies who can own which class of stock, what each class can do, and what happens if anyone tries to break the loop. After reading it, you realize this is not a provision about control. It is a complete design for how control passes between generations, disappears automatically upon transfer, and reinforces itself against external challenge.
SpaceX will have three classes of common stock after it goes public.
Class A is what you can buy in the IPO. One vote per share. Everything traded on the open market is Class A.
Class B carries supervoting rights. Ten votes per share. Musk and his affiliated entities hold nearly all of it. More importantly, the S-1 states explicitly that any future issuance of Class B shares goes only to Musk, his family members, and specific permitted entities defined in the charter. The public market can never get Class B.
Class C has no voting rights. Zero votes. It has not been issued yet, but the charter authorizes 10 billion shares, mainly for future employee equity or acquisition consideration. It dilutes economic interest without diluting voting power.
So far, this is a conventional dual-class design. Facebook used a similar structure when it went public in 2012 (Zuckerberg retained control through Class B). Google did the same in 2004. Multi-class stock is not new.
But keep reading. The conventional part stops here.
The standard practice in dual-class structures is this: supervoting shares automatically convert to single-vote shares when transferred. This prevents supervoting power from spreading beyond the founder’s control. Facebook has this rule. Google has it too.
SpaceX has a similar rule: Class B shares automatically convert to Class A upon a “Transfer.” So far, nothing unusual.
But the definition of “Transfer” and the exceptions are where the real architecture lives. The S-1 defines Transfer to cover nearly every conceivable way to move economic interest: sale, assignment, pledge, gift, encumbrance, grant of proxy, whether for value or not, voluntary or involuntary, by operation of law or otherwise.
Then comes a long list of exceptions. Permitted transfers include: transfers to Musk or his permitted entities, transfers to Musk’s family members, transfers to trusts, partnerships, or LLCs controlled by Musk or his family members, transfers to certain charitable trusts, transfers to retirement accounts. To summarize: as long as the recipient is inside Musk’s control circle or bloodline, Class B stays Class B. Step outside that circle, and Class B becomes Class A on the spot.
To see what this means, take a concrete scenario. Suppose you are an early SpaceX employee holding some Class B shares. You want to sell. You find a buyer, agree on a price, and close the deal. At the moment of closing, those shares convert from Class B to Class A. The buyer gets single-vote stock. He pays one price and receives something different. You are not selling supervoting rights. You are selling an asset that automatically downgrades to single-vote status the instant the transaction completes.
Now look at it from Musk’s side. His Class B shares can move through a carefully designed set of permitted transfer paths without triggering conversion. From his name to a trust, from the trust to a family limited partnership, from one generation to the next. The assets move. The supervoting power does not leak.
This is what the structure actually does. It makes supervoting rights a non-tradeable, non-transferable, non-priceable asset. It belongs to one person and his designated successors. You can own the economic interest in SpaceX. But you can never buy even a sliver of voice in how the company is run, not because anyone stops you, but because the right itself has been engineered to be untradeable.
Back to the quote at the top. Musk can only be removed as CEO and Chairman by a vote of Class B shareholders. Class A shareholders have no say.
Who holds Class B? Musk does. A few early insiders might hold some too. But the S-1’s beneficial ownership table shows that Musk, through direct holdings and trusts, controls the vast majority of Class B voting power.
The logic chain goes like this. You want to remove Musk. You need a majority of Class B shareholders to approve. Musk controls the majority of Class B voting power. Musk would need to vote to remove himself.
The S-1 never states this logic explicitly. But it lays out every link in the chain clearly enough that any reader can connect them and arrive at the same conclusion. That approach is more powerful than simply writing “Musk cannot be removed.” A statement can be amended. A mechanism is embedded across the charter, the cap table, and corporate law. Changing it means changing every piece at once.
Some might ask: what about the board? Normally, even when a founder holds supervoting rights, removing the CEO remains a board function. The founder can appoint directors and control the board’s composition, but the act of removing a CEO nominally belongs to the board. SpaceX handles this differently. It takes that power away from the board entirely and writes it into the charter, assigning it to a voting group. That voting group is Musk himself.
Harvard Law professor Lucian Bebchuk said something about this in a Reuters interview: “This provision is not common.” When a Harvard professor who specializes in corporate governance says “not common,” you can read that as academia’s way of saying “never seen before.”
There is another variable here: the state of incorporation. SpaceX is incorporated in Texas.
Most US public companies are incorporated in Delaware. Not because Delaware corporate law is “better,” but because Delaware has the deepest body of case law. Nearly every governance dispute has precedent. The judges are specialized. Decisions are predictable. Uncertainty is expensive, so companies pay Delaware’s franchise tax in exchange for predictable legal outcomes.
SpaceX moved from Delaware to Texas in 2024, following the same path Tesla took. The motivation is not hard to understand. A Delaware court voided Musk’s $56 billion Tesla compensation package in 2024 (later reinstated by the Delaware Supreme Court). For someone who views founder control as a core governance design, Delaware is a legal risk.
Texas corporate law (the Texas Business Organizations Code) offers more flexibility around supervoting rights, classified boards, and similar arrangements. Its case law is far less developed than Delaware’s. That means when a governance dispute does arise, the legal outcome is less predictable. And who benefits from low predictability? The incumbent controller. Challengers face a legal environment with no clear precedent. Litigation costs and uncertainty reinforce the controller’s position.
The S-1 acknowledges this itself. In the Anti-takeover section of Capital Stock, it notes that Section 21.606 of the Texas TBOC imposes a three-year moratorium on business combinations between an “affiliated shareholder” and the company. The definition of “affiliated shareholder” includes Musk. In other words, even if someone accumulates a large block of Class A shares on the open market, they cannot push through any control transaction requiring shareholder approval for three years. And even after three years, Class B shareholders can block it unilaterally.
You get the economic interest in a company. This company generated $18.7 billion in revenue in 2025, of which Starlink contributed $11.4 billion. It has 10.3 million paying broadband subscribers across 164 countries. It has launched more reusable rockets than any organization in human history. Its operational performance is real. Its technological moat was built over two decades of engineering.
What you do not get is any meaningful participation in how the company is governed. You cannot vote to remove the CEO. You cannot elect a majority of the board (Class B separately elects 51% of the directors). You cannot push through charter amendments (any amendment affecting Class B rights requires Class B approval alone). You cannot submit a shareholder proposal unless you hold at least $1 million in market value or 3% of the voting power, have held it for at least six months, and can solicit support from at least 67% of the voting power.
Each of these restrictions has precedent on its own. Facebook’s Zuckerberg has supervoting rights. Google’s founders do too. Snap went public in 2017 with stock that carried no voting rights at all.
But SpaceX does not pick one of these mechanisms. It uses all of them at once. Class B has a fixed majority on the board. CEO removal is decided by Class B alone. Class B is issued only to one person and his blood relatives. Transfer triggers automatic conversion. Texas incorporation. Anti-takeover provisions. Impossibly high thresholds for shareholder proposals. Every item reinforces the same outcome, and these mechanisms lock each other in place. To change any one of them, you first need to get past another mechanism’s control. This structure is not a wall. It is a courtyard with no gate.
From 2022 to 2026, the US IPO market was effectively closed. There were only 181 IPOs in all of 2022 (down from 1,035 in 2021), and the number shrank further in 2023. The market was waiting for a signal. SpaceX’s IPO was expected to be that signal. Its scale, brand, and technology narrative were big enough to reopen the IPO window.
But the signal this S-1 sends is different from what people expected. It tells the market: you can participate in the largest IPO in human history, but you will have no say in how this company runs. You are not buying a share in a company in the traditional sense. You are buying a new asset class. It gives you economic exposure but no governance rights. It is called common stock, but it is no longer the same thing as the common stock of the past.
Will this affect demand? Probably not. SpaceX is a scarce asset, and scarce assets do not need to offer full rights.
But the longer-term impact is bigger. The IPO market is shifting from a two-sided negotiation (both issuer and investor set terms) to a seller-priced market. There are too few good companies and too much capital chasing them. When the supply side does not need to compromise, governance terms will keep moving in the direction that favors the controller. SpaceX has simply pushed this trend to its current endpoint. When the next hot unicorn goes public, its lawyers will open SpaceX’s S-1, point to the Capital Stock chapter, and say: “This is how far they got last time. Can we push it further?”
This article cannot answer how far is too far. But it can tell you that the final step is no longer far away.
This article is based on the Capital Stock, Risk Factors, Security Ownership, and Related Person Transactions sections of SpaceX’s S-1 registration statement (CIK 0001181412), filed with the SEC on May 20, 2026.