SpaceX, OpenAI, and Anthropic Are Going Public Because AI Is Burning Cash Faster Than Private Markets Can Supply It

The Numbers First

SpaceX raised $75 billion when it went public, according to The Atlantic, instantly landing among the ten most valuable companies in the world. OpenAI and Anthropic have both recently filed to go public later this year. If those offerings close, we will likely have witnessed the three largest IPOs in history within a single calendar year.

This is not a typical pattern, and the reason is straightforward.

AI Costs an Absurd Amount of Money

SpaceX's AI business is burning through $1 billion every month, according to The Atlantic. OpenAI lost a reported $38.5 billion last year, according to financial statements uncovered by blogger Ed Zitron. These companies are consuming capital at a scale that private venture markets simply cannot sustain indefinitely.

This is the core reason for the IPO wave. Private markets alone cannot provide the tens of billions of dollars needed to build out data centers and stay competitive in a race where falling behind could mean irrelevance.

How This Differs From the Last Tech Wave

Facebook, Uber, Airbnb, and most of the previous generation of tech giants went public largely to let early investors and employees sell shares. Their underlying businesses did not require massive up-front capital. Software companies could build at relatively low cost, reach profitability, and then use the IPO as an exit mechanism.

AI infrastructure works differently. Building and running large language models requires physical hardware, enormous power consumption, and continuous reinvestment. The Atlantic's framing here is accurate: this is a structural change in the tech industry's capital model, not just a cyclical IPO boom.

Since 2003, public companies as a group have returned more money to shareholders than they raised from new stock issuances. That net-return dynamic now appears to be reversing, with major AI companies drawing fresh capital from public markets at scale. Alphabet recently announced it would sell new stock to raise $85 billion, all earmarked for AI. Oracle said it would sell $20 billion in stock. Meta is reportedly contemplating a big stock offering as well. JPMorgan Chase estimates that companies will sell $1.5 trillion in shares over the next two years.

The Shrinking IPO Market as Context

The broader IPO market has actually collapsed over the past three decades. According to The Atlantic, roughly 500 companies per year went public in the 1990s. That number has dropped to about 120 per year over the last decade. The total count of U.S. public companies is down nearly 40 percent from its 1990s peak, even as the number of new businesses being formed has surged.

Venture capital and private equity filled that gap for most companies. But AI's capital requirements have outgrown what private markets are willing or able to provide on their own.

The Strongest Counterargument

Skeptics have a legitimate concern: these companies are not profitable, may not be profitable for years, and are asking public investors to fund losses measured in the tens of billions annually. The historical comparison to the dot-com bubble is not frivolous. As economist and hedge-fund manager Owen Lamont has written, according to The Atlantic, "When firms are selling, you should generally sell as well." Historically, a big surge in stock supply overwhelms demand and raises questions about whether markets are properly reflecting value or inflating a bubble.

The volatility of SpaceX stock, which has tumbled for days after its IPO erasing nearly all gains enjoyed by the average investor, shows just how much uncertainty attends even the most hyped offerings in a risky industry.

What Investors Are Actually Betting On

Public investors buying into these IPOs are not buying current earnings. They are buying a thesis: that whoever wins the AI infrastructure race will generate outsized returns for decades. That thesis could be correct. It could also be wrong.

The Atlantic's own conclusion is blunt: these IPOs and stock sales are happening because companies are trying to raise as much money as possible while the good times last. That, The Atlantic notes, should make observers wonder how much longer the good times will last.

The Open Question

OpenAI and Anthropic have both filed to go public later this year, but have not yet priced their offerings. The actual valuations at which they price their offerings, and whether institutional demand matches current expectations, will determine whether this IPO wave validates the AI investment thesis or marks the peak of it.