infrastructureScore 70/100Research
Nemotron 3 Ultra's 550B parameter size requires hundreds of gigabytes of GPU memory, driving demand for cloud GPU infrastructure like Lambda
Andrej Karpathy· Independent· AI· 2026-06-14· about Lambda Labs
“However, no. Because I would love to run it locally too. But it's huge. 550 billion parameters. You need hundreds of gigabytes of GPU memory for that. This is why I will probably use it on Lambda.”
Why it matters
The massive parameter count of open-source models like Nemotron forces users toward cloud GPU providers, creating sustained demand for inference infrastructure that cannot be served by consumer hardware.
Investment implication
Cloud GPU providers (Lambda, RunwayML, others) and GPU cloud platforms benefit from the trend toward larger open-source models that cannot run locally. This is a pick-and-shovel opportunity in the open AI ecosystem.