SpaceX and OpenAI Want to Be Worth More Than Berkshire Hathaway. Heres What They Are Not Telling You.

SpaceX and OpenAI Are About to Make Warren Buffett Look Small. Can They Answer the One Question That Matters?

In the summer of 2026, if the bankers are right, the world's most valuable companies will no longer be insurance conglomerates or consumer brands. They will be a rocket company that loses $4 billion a quarter, and an AI lab that burns cash at a pace its own executives struggle to explain. SpaceX filed its S-1 last week targeting a $1.75 trillion valuation. OpenAI is expected to file its own paperwork as early as Friday, targeting a valuation of up to $1 trillion, per the Wall Street Journal. Together with Anthropic — which is also widely expected to go public this year — these three companies could enter public markets with a combined valuation north of $2.9 trillion. That would make them worth more than Berkshire Hathaway, the 60-year-old holding company that has spent decades accumulating railroads, insurance float, and household names into the world's most revered investment vehicle. It would also require public markets to absorb somewhere between $432 billion and $576 billion in new stock in a single quarter — roughly equal to every IPO the United States produced across the entire decade from 2016 to 2025.

The numbers are real, and they are being greeted with a kind of awe that borders on reverence. Polymarket traders are betting that all three companies will debut above $1 trillion on their first day of trading. The underwriting banks — Goldman Sachs and Morgan Stanley for SpaceX — are reportedly already fielding calls from institutional investors desperate for allocation. This is, by any measure, the most consequential week in capital markets since the dot-com era, except this time the companies being crowned champions before a single share trades publicly.

The valuations themselves rest on assumptions that would have seemed fantastical even five years ago. SpaceX generated $4.69 billion in the first quarter of 2026, up 15 percent year-over-year — impressive until you note that its growth rate was more than halved from the 33 percent it delivered in 2025. Its net loss in the same quarter was $4.28 billion, nearly matching its loss for all of 2025. The company is burning faster than it is growing. OpenAI is generating $25 billion in annualized revenue, per filings cited by multiple outlets, but is understood to be operating at a loss as it invests aggressively in model development and compute infrastructure.

The financial architecture beneath these valuations is, at minimum, unusual. SpaceX's S-1 discloses a contract worth $1.25 billion per month through May 2029 — $15 billion per year — with Anthropic for compute infrastructure. In the first quarter of 2026, SpaceX generated $4.69 billion in total revenue. The Anthropic deal alone would triple that figure annualized. Seeking Alpha's analysis of the filing frames it as high-margin revenue for SpaceX: the compute arrangement is structured as Anthropic purchasing access to SpaceX's Colossus supercomputer cluster. For public shareholders, this creates a distinct governance problem. Elon Musk controls SpaceX; he also controls xAI, which competes directly with Anthropic. The arrangement means SpaceX's board is obligated to maximize returns from a counterparty that is also Musk's AI rival — a tension the S-1 does not resolve.

This is not necessarily illegal. It may be entirely commercial. But it is the kind of detail that public markets are designed to surface, and that private markets have historically buried.

The capital absorption problem is similarly without precedent. According to Tomasz Tunguz, a former Google and Redpoint executive who writes about venture and public markets, the three IPOs would require public markets to absorb between $432 billion and $576 billion at standard float percentages — meaning the percentage of total shares made available to public investors on day one. For context, the entire U.S. IPO market raised $469 billion across the decade from 2016 through 2025. In a single quarter, public investors would be asked to fund roughly a decade's worth of new listings, concentrated in three companies that have never faced the scrutiny of a public earnings call, an activist short-seller, or a pension fund with a governance committee.

SpaceX's Starlink division, which provides satellite internet, generated $3.25 billion in the first quarter of 2025, up 27 percent year-over-year, and is the only part of the company producing positive adjusted EBITDA. The rest of SpaceX — Starship development, the xAI compute payments, the various moonshots — is subsidized by Starlink's cash flow and by the confidence of private investors who have ridden Elon Musk's track record to returns that have no visible ceiling. OpenAI's revenue growth is real, but its cost structure is opaque, and its path to profitability depends on assumptions about AI infrastructure scaling costs that its own researchers have described as uncertain.

What none of this establishes is whether the companies themselves are worth $1.75 trillion, or $1 trillion, or $380 billion. What the filings actually demonstrate is that private markets have constructed a set of valuations that public markets will now be asked to validate — or reject. Warren Buffett spent 60 years building Berkshire Hathaway through disciplined capital allocation, insurance float arbitrage, and a refusal to pay for narrative. SpaceX and OpenAI are being valued, in significant part, on the basis of what they might become.

The question that public markets will eventually ask — and that the first earnings call, the first activist investor, the first quarter of disappointing guidance will force — is simpler and older than any of the technology involved: What do you actually earn? SpaceX and OpenAI will have to answer that question in public, for the first time, starting sometime this fall. The rest of us will find out whether the answer is worth $2.9 trillion.