Burry is not saying AI is fake. He is saying the price is wrong.

This distinction matters because every time Burry warns about a bubble, the response is the same: "But the earnings are real." They were real in 2000 too. Cisco had record revenue. Intel was printing money. Sun Microsystems was growing 30% a year. The earnings were real. The multiples were not sustainable.

AMD just reported 38% revenue growth. Nvidia is expected to show similar numbers on May 20. The demand for AI chips is genuine. Lisa Su described "tremendous demand" from agentic AI workloads. None of that is in dispute.

What Burry is disputing is the price you pay for that growth. SOX at a 25-year high with a 42% single-month gain means every dollar of future earnings growth is already priced in, plus some. If growth merely meets expectations instead of beating them, the stocks go down. If growth decelerates from 38% to 25%, the stocks crash. Thats the math of extreme multiples.

The 2000 comparison has a specific structural parallel

Burry isnt just pointing at a chart shape. The parallel he is drawing has a mechanical basis. In 1999 to 2000, semiconductor stocks surged because every company on earth was buying hardware for the internet buildout. The demand was real. What wasnt real was the assumption that the buying rate would continue at peak levels indefinitely.

The same dynamic is happening now. Hyperscalers are spending record amounts on AI infrastructure. Microsoft, Google, Amazon, and Meta are all in a capex arms race. That spending is driving chip revenue. But capex is front-loaded by nature. You build the data center once. You dont build it again next quarter.

Burry is betting that the market is extrapolating peak capex as if its permanent. The moment any hyperscaler signals a slowdown in AI spending, even a deceleration from 40% growth to 30%, the semiconductor sector reprices the entire forward earnings curve in a single session.

Paul Tudor Jones agrees on the shape but disagrees on the timing

Jones told CNBC this week that the current environment "feels similar to 1999, roughly a year before technology shares peaked." He estimated the rally could continue for another year or two before it ends.

That distinction is everything. Burry is buying January 2027 puts, roughly eight months out. Jones thinks you might have twelve to twenty-four months. Both agree the cycle ends. They disagree on when.

For investors, this means the question isnt whether the semiconductor rally is sustainable forever. Nobody serious thinks it is. The question is whether you can ride the remaining upside without getting caught in the reversal. Burrys track record says the reversal comes sooner than you expect. Joness track record says you leave money on the table if you exit too early.