Jensen Huang explains why Nvidia will grow an astounding 70% next year

Jensen Huang spent Thursday at Goldman Sachs’ Communacopia + Technology conference making the case that Nvidia’s AI boom still has room to run. The message was simple: the company that people still describe as a chip maker is, in his view, much closer to the plumbing of the AI industry than that label suggests.

Huang leaned hard on scale. He said one Nvidia system is no longer a $399 graphics card but an $8.5 million setup made from 36 Grace CPUs and 72 Blackwell GPUs, tied together with NVLink and built from 2 million parts. He also said orders for that system are growing 27% month to month.

The bigger claim was about next year. Huang repeated Nvidia’s forecast that revenue could rise 70% year over year, after analysts expect the current fiscal year to end at about $400 billion. On that math, next year would land around $680 billion. That is an enormous number, and Huang argued it is plausible because Nvidia runs every model and serves every major kind of AI customer, from Anthropic and OpenAI to Google and open-weight models.

He also said Nvidia has visibility across the supply chain, from memory chip makers to data center projects, and claimed the company is tracking “every single gigawatt of land, power, shell around the world.” That confidence is what lets him shrug off worries about rivals, including hyperscalers building their own chips, AI labs doing the same, Cerebras, and startups like Etched.

The circular-deal chatter got the usual Huang treatment: a joke. He said Nvidia puts in $1 and gets $100 back, then added that the company only invests after seeing real contracts. He said he has seen $100 billion worth of such contracts. The larger question is whether this really keeps going forever. Huang clearly thinks Nvidia still has the best seat in the house.