So everyone's talking about michael burry again and his whole "AI is the next dot-com bust" thing. Look, I get it - the guy called 2008 perfectly and made a fortune doing it. That's legendary status cemented. But here's the thing: his recent track record has been... rough. Since that 2008 windfall, he's been consistently early and wrong on his bearish calls while markets kept climbing. He even shut down his hedge fund last year because he was too misaligned with how things were actually playing out.

Now michael burry is making noise about AI stocks being in a 1999-style mania that'll crash just like the dot-com bust. The thesis sounds compelling on the surface, but when you dig into the actual numbers, it falls apart pretty quickly.

His main argument is that big tech is cooking the books with depreciation schedules to pump their earnings. Meta, Microsoft, Alphabet - they're all allegedly using artificially long depreciation timelines on servers and GPUs. Here's why that doesn't hold up: yeah, newer GPUs get released constantly, but older chips don't become worthless. They run inference - that's the actual revenue-generating part where AI companies make money. Plus most AI infrastructure actually has a 15-20 year useful life, so the depreciation schedules aren't that aggressive when you look at the full picture.

Then michael burry warns about massive cash flow strain from all this CAPEX spending. Except the data shows the opposite. Alphabet's operating cash flow jumped from under $100 billion to $164 billion recently. Margins are expanding across the board. And get this - companies running AI are reporting returns of $3+ for every $1 they invest. The new wave of agentic AI is supposedly cutting costs by 25% or more. That's not a sign of a bubble, that's actual value creation.

He also compares NVIDIA to Cisco in 2000, suggesting NVDA is overvalued just like CSCO was. But the comparison is weak. Cisco's P/E hit over 200 when it peaked in March 2000. NVIDIA's current P/E is sitting at 47. Totally different ballpark.

What's actually interesting right now? H100 GPU rental prices have jumped about 17% since mid-December. That's a signal of real, sustained demand - not hype. Companies like Nebius and CoreWeave are positioned well for that. And the energy constraint is becoming the real bottleneck, which is why Bloom Energy is getting attention.

Options traders are clearly betting bullish too. Someone dropped nearly $9 million on March $205 calls for NVIDIA ahead of earnings. Bloom saw some massive call buying as well. That's smart money putting real capital behind their conviction.

Look, michael burry's contrarian instincts made him a fortune in 2008, but being a contrarian doesn't mean you're always right. The AI infrastructure story has actual fundamentals backing it up - cash flow, margins, ROI metrics, real cost savings. That's different from 1999 when it was all eyeballs and hope. The data just doesn't support the bubble narrative he's pushing.