Three companies are preparing to deliver the largest wave of initial public offerings in American history. SpaceX, OpenAI, and Anthropic have a combined implied valuation approaching $3 trillion. Not one of them is profitable.

Reuters reported on Thursday that there is no precedent for this in U.S. market history: three unprofitable companies of this scale going public in the same window. The combined offering would dwarf anything seen during the dot-com boom, the 2021 SPAC frenzy, or the Facebook-era IPO cycle.

The numbers are staggering. OpenAI closed a $122 billion funding round, the largest venture deal in history, at an $852 billion valuation. It has 810 million monthly active users, over $20 billion in annualized revenue, and one million enterprise customers. It just shipped GPT-5.5. It is targeting a listing as early as Q4 2026.

Anthropic's trajectory is equally extreme. Revenue tripled from $9 billion to $30 billion in four months. Its secondary market valuation hit $1 trillion last week, surpassing OpenAI on Forge Global's platform. Google committed $40 billion on Friday. Amazon has committed $33 billion. It has $73 billion in hyperscaler backing and is reportedly considering an October IPO. Goldman Sachs, JPMorgan, and Morgan Stanley are all in the room.

SpaceX, the third leg of the triangle, adds another dimension entirely. Musk's rocket company has been the most sought-after private company on secondary markets for years. Now, with the Cursor $60 billion acquisition option and xAI integrated into its capital structure, its AI surface area has expanded dramatically.

The Profit Problem

Here is where it gets uncomfortable. Both OpenAI and Anthropic are reporting year-over-year revenue growth of roughly 10x for three consecutive years. That rate of scaling has no historical precedent. But the cost side is where the IPO thesis gets complicated.

OpenAI's model training costs are reportedly projected to rise from $30 billion in 2026 to $121 billion by 2028. That consumes all projected revenue in the interim. Anthropic's training costs are projected at $42 billion by 2029. The revenue curves are extraordinary. The cost curves are more extraordinary.

This creates a dynamic that public market investors have never been asked to underwrite at this scale: companies whose revenue is growing at the fastest rate in corporate history, whose costs are growing even faster, seeking valuations that assume both problems resolve simultaneously.

The First Quarter Told the Story

Q1 2026 saw somewhere between $297 billion and $314 billion invested globally in venture capital. AI accounted for 80% to 81% of all venture dollars. The four largest venture deals in history all closed in the same quarter: OpenAI at $122 billion, Anthropic at $30 billion, xAI at $20 billion, and Waymo at $16 billion.

We reported last month that four companies took 65% of all global venture capital in a single quarter. That concentration has only intensified. The venture ecosystem is not diversifying. It is consolidating around a bet that these specific companies will define the next computing platform.

The Valuation Gap

There is already a massive disconnect between what secondary markets are pricing and what IPO bankers expect to sell. Anthropic trades at $1 trillion on Forge Global. Its IPO bankers are reportedly targeting $400 billion to $500 billion. That is a 2x gap. Someone is going to be wrong.

OpenAI's situation is similar. Some investors are questioning whether the IPO would need to exceed $1.2 trillion to justify current private market investments. That would make it the largest technology IPO ever attempted, by a company that is actively losing money.

What This Means

The AI IPO wave is not a bubble thesis or a bull thesis. It is a structural question about how capital markets price a technology transition that is happening faster than any valuation framework was designed to handle.

Amazon lost money for nearly two decades before becoming the most valuable company on earth. Netflix hemorrhaged cash while building the streaming category. But neither Amazon nor Netflix were valued at $1 trillion before going public. Neither required $100 billion in annual training costs just to maintain competitive parity.

The bet is that AI is different. That the revenue curves will outrun the cost curves. That the market is right to price these companies at generational levels before they have generated a single dollar of profit.

If that bet is right, Q4 2026 will be remembered as the moment the AI era went public. If it is wrong, it will be remembered as the moment the AI era got repriced.

First reported by Reuters, with additional reporting from TechCrunch and InvestorPlace.