Paul Tudor Jones, the billionaire hedge fund manager who famously predicted the 1987 crash, went on CNBC's Squawk Box on Wednesday and said something that should make every AI investor pay very close attention.

The AI bull market, he said, "has another year or two to run." He compared the current moment to 1999, roughly a year before dot-com share prices peaked in early 2000. He drew parallels to Microsoft's early software dominance in the 1980s and the commercialization of the internet in the mid-1990s.

And then, in the same breath, he said he recently bought more AI stocks.

Let that sit for a second. One of the most respected investors alive is telling you the party ends in a year or two, that the drawdown when it comes could be significant, and that he is still buying. That is not a contradiction. That is the playbook of someone who has ridden bubbles before and knows exactly when to get out.

The 1999 Parallel Is More Precise Than You Think

Jones specifically mentioned ChatGPT and Claude Code as his Apple II moment. Just as the Apple II in 1977 showed people what personal computing could actually do, Claude Code and its competitors crossed an invisible threshold in late 2025 where AI went from interesting gadget to genuinely transformative technology. Developers report 4x productivity gains. Uber burned its entire 2026 AI coding budget in four months because adoption was too fast. Meta laid off 10% of its workforce because, as Zuckerberg told investors, AI lets one person do the work of entire teams.

Jones sees this as the real thing, not just hype. The revenues are actually showing up. Anthropic is growing faster than Zoom did during the pandemic, faster than Google in the early 2000s, faster than Standard Oil in the Gilded Age. The Atlantic published a piece this week titled "So, About That AI Bubble" that argued the bubble thesis is getting harder to defend now that Claude Code has turned developer productivity from a promise into a metric.

The Crash Warning Nobody Wants to Hear

Here is the part people will ignore: Jones also said that when the AI bull market does end, the drawdown "could be significant." He is not saying AI is fake. He is saying the market pricing of AI is getting ahead of the revenue reality, just like it did with the internet. The internet was real. The companies were real. The crash was also real. Amazon went from $107 to $7. It took 10 years to recover.

The difference between 1999 and now, according to Jones, is that AI adoption is being driven by real productivity gains rather than speculative promise alone. But he is explicitly framing this as a timing call, not a fundamentals call. The music is still playing. The question is whether you will hear it stop.

What This Means for You

Jones' track record demands attention. This is the man who shorted the market before Black Monday in 1987 and turned $7.5 billion into what is now one of the most respected macro funds in history. When he says "another year or two," that is not a guess. That is a framework honed over four decades of navigating exactly these kinds of cycles.

The takeaway is not "sell everything." It is not even "be cautious." It is: know where you are in the cycle. Jones knows. He is still buying because the upside in the next 12 to 24 months is worth the eventual crash. But he will not be the last one out the door. That is the difference between a billionaire and a retail investor reading this article. He has a plan for the exit. Do you?

Jones' full interview aired on CNBC's Squawk Box. The Atlantic's analysis of AI revenue growth is here.