Just been diving into Dr Michael Burry's latest AI thesis and honestly, it's not sitting right with me. The guy's calling this a dot-com 2.0 situation, claiming we're in a 1999-style mania that's bound to crash. On paper it sounds compelling - he nailed the 2008 crisis after all. But when you actually look at the numbers? The thesis starts falling apart pretty quickly.

So here's what Dr Michael Burry is arguing. First, he's saying the big tech companies are cooking the books with sketchy depreciation schedules. He points to how Alphabet depreciates servers over just 4-6 years to pump up earnings. Sounds damning until you realize most AI infrastructure actually lasts 15-20 years. Plus, older GPUs don't just become worthless - they're still valuable for inference work. That's a pretty significant hole in his logic.

Second claim: the massive CAPEX spending is going to strangle cash flow. Burry's worried these companies are overinvesting without returns. But that's where the data completely contradicts him. Alphabet's operating cash flow jumped from under $100B to $164B in 2026. Margins are expanding across the board. And get this - companies scaling agentic AI are reporting returns of more than $3 for every $1 invested. That's not a cash flow problem, that's a cash flow bonanza.

Then there's the NVIDIA comparison. Burry's drawing parallels to Cisco topping in 2000, citing valuation concerns. Except when you actually compare the numbers, it falls apart. Cisco had a P/E above 200 when it peaked in March 2000. NVIDIA's trading at a P/E of 47 right now. That's not even close to the same situation.

What's actually interesting is what the market's telling us right now. H100 GPU rental prices have spiked roughly 17% since mid-December. That's a pretty clear signal of continued scarcity and demand. The agentic AI wave is driving this, and it's benefiting infrastructure plays like Nebius, CoreWeave, and IREN. Energy companies like Bloom Energy are also getting attention because power consumption remains the real bottleneck for hyperscalers.

Speaking of Bloom, options traders have been making some seriously bullish bets. Monday alone saw a million-dollar call position in Bloom and a whale dropped nearly $9 million on March $205 calls for NVIDIA ahead of earnings. Bloom shares jumped 8% that day, actually bucking Nasdaq weakness and setting up a textbook bull flag on the weekly chart.

Look, I respect Dr Michael Burry's track record. The contrarian approach that worked in 2008 is part of his legend. But right now? The AI narrative he's bearish on is getting backed up by actual operational data, not hype. H100 scarcity, expanding margins, positive cash flow dynamics - this doesn't look like dot-com 2.0. It looks like infrastructure that's actually generating returns. That's a fundamentally different setup.