"I stand by my analysis. I am not claiming Nvidia is Enron.
"Michael Burry · investor · 2026
What people mean by the AI bubble
In 2026 the phrase points to one worry: the money going into AI infrastructure may be running far ahead of the money coming back out of it. The big cloud players are spending well over 600 billion dollars a year on chips and data centers, while some of the largest deals loop back on themselves, as Nvidia pledges to invest in OpenAI, OpenAI rents compute, and that spend flows back to the same suppliers. Bulls see a genuine platform shift with fast growing revenue and demand that outruns supply. Bears see circular financing, a wide gap between spend and return, and chips that depreciate faster than the books assume.
2026 hyperscaler AI capex
over $600B
OpenAI annualized revenue
over $20B
Enterprise GenAI pilots with no measurable return
95%
GPU useful life, reported vs bear estimate
6 yr vs ~3 yr
Is this a bubble about to pop, or the early innings of a real platform shift?
Make the strongest case for your side, name the one number that convinces you, and see where others land.
Bull: OpenAI cleared 20 billion dollars annualized and clouds report demand outrunning supply.
Bear: capex tops 600 billion a year while an MIT study found 95 percent of AI pilots show no return.