Nvidia’s $100bn OpenAI "investment" was always a bad joke
Surprise, surprise. Nvidia is officially walking back the massive $100 billion investment in OpenAI they touted last year. You know the one. It was the ultimate circular Ponzi scheme: Nvidia gives OpenAI cash, OpenAI hands it right back to buy Nvidia chips, and Nvidia books it as “revenue” to keep their stock chart looking like a vertical line.
It was a desperate attempt to pretend the demand for GenAI was still going to the moon. But by late 2025, reality hit hard. The “anti-GenAI” fatigue set in, the energy grid started screaming, and the world realised that endless chatbots aren't actually solving expensive problems.
Jensen sees the writing on the wall:
Jensen Huang isn't stupid. He knows the difference between a gold rush and a bubble. When the guy selling the pickaxes refuses to lend money to the miners, you know the mine is empty. He looked at OpenAI burning $17 billion a year with no clear path to profit and decided that $100bil figure was “no, nothing like that.” It’s just going to be “huge” instead.
The other idiots in the room:
If the biggest beneficiary of this bubble (Nvidia) realises their biggest user (OpenAI) isn't worth the financial risk, where does that leave the rest of them? I’m looking at Microsoft, Meta, Amazon and the gang. They are still blindly pouring billions into data centres, seemingly oblivious to the fact that the hardware supplier itself is getting cold feet.
✅ The Verdict
If the poster boy and de facto hardware supplier of the “AI revolution” is hesitating to spend their own cash on “AI”, maybe, just maybe, the other big boys should take notes. Obviously they won't, considering the level of corporate brain rot involved, but at least Nvidia was smart enough to jump off the train before it derailed.