The AI boom has been built on one assumption.
Spend first.
Figure out the profits later.
That works only if there’s always another funding round… or eventually an IPO.
Now that assumption is starting to look a little less certain.
There’s growing discussion that an Anthropic IPO this year has become much less likely after a rough month for AI assets, continued legal battles, questions around profitability, and signs that public market appetite has cooled. Anthropic itself has maintained that any listing depends on market conditions.
By itself, that’s just one company.
But it fits a much bigger pattern.
SpaceX has reportedly traded below recent private valuations.
Several AI names have postponed IPO plans.
Long-term Treasury yields have climbed to levels not seen since 2007, making capital more expensive.
Investors are no longer rewarding every AI story simply because it has “AI” attached to it.
Amazon actually proved the opposite this week.
It spent an astonishing $220 billion, burned cash, and the stock still rallied because AWS delivered 37% growth.
That tells you something important.
Wall Street isn’t done funding AI.
It’s done funding AI without proof.
That’s a huge change.
Think about the entire AI ecosystem.
Most frontier labs are still burning enormous amounts of cash.
Data centers keep getting bigger.
Power demand keeps climbing.
HBM memory remains constrained.
The bond market keeps raising the cost of capital.
Meanwhile, exit opportunities are becoming less certain.
That’s a dangerous combination for venture investors.
If IPOs slow down, venture funds can’t return capital.
If they can’t return capital, raising the next fund becomes harder.
If the next fund gets smaller, financing the next generation of AI companies gets harder too.
That’s how liquidity cycles reverse.
Not because AI suddenly stops improving.
Because the financial machine supporting it starts slowing down.
The irony is that AI may have solved the technology problem before it solved the business problem.
Everyone knows how to build increasingly capable models.
Far fewer have shown they can generate enough cash to justify hundreds of billions in infrastructure spending.
That’s why I don’t think the real story is whether Anthropic lists this quarter or next.
The real story is that the market is beginning to ask a question it avoided for two years.
Where are the exits?
When money was free, nobody cared.
With long-term yields back near 2007 highs and investors demanding profits instead of promises, that question suddenly matters a lot more.