Nvidia CEO reiterates 70% revenue growth outlook for next year
Nvidia CEO Jensen Huang has reiterated that the company could increase revenue by 70% year on year next year, extending a record-breaking run in AI infrastructure sales. Speaking at the Goldman Sachs Communicopia + Technology conference, Huang said analysts expect Nvidia to finish its current fiscal year at roughly $400 billion in revenue, implying about $680 billion if that growth rate is achieved.
Huang’s case rests on Nvidia’s position across the AI ecosystem, from component suppliers and system builders to cloud operators, AI laboratories and data-centre projects. He said the company’s platform runs models from Anthropic, OpenAI and Google as well as open-weight models, describing Nvidia as a foundational platform for the AI industry.
From individual chips to large-scale systems
Huang argued that the market still understates the scale of what Nvidia now sells by viewing the company principally as a chip supplier. He contrasted an earlier consumer GPU price of $399 with a current system he called a GPU: an NVLink-connected installation valued at $8.5 million, comprising 2 million parts and requiring 250,000 kilowatts of power. Nvidia ships thousands of such systems, he said.
He singled out a computer system containing 36 Grace CPUs and 72 Blackwell GPUs, saying orders for that product are rising 27% month to month. The example illustrates why the company’s revenue outlook is tied to integrated computing installations and the facilities needed to deploy them, rather than to standalone accelerator shipments alone.
Visibility across AI infrastructure demand
Nvidia is tracking data-centre development through land, power and building shells, Huang said, alongside information from neoclouds, OEMs, cloud providers and AI-native companies. In this context, Nvidia record revenue and startup investment activity frames Nvidia’s recent record revenue and startup investment activity as part of the broader infrastructure build-out Huang described.
The CEO addressed questions about investments in companies that subsequently purchase Nvidia equipment. He rejected the description of those arrangements as circular, saying Nvidia puts in relatively small amounts while receiving much larger sales. More substantively, he said Nvidia requires evidence of real customer contracts producing revenue before making an investment, and said he had seen $100 billion in such contracts.
Competition and execution remain central
Nvidia faces growing AI-chip competition from hyperscalers including Amazon, Microsoft and Google, which are developing their own hardware, as well as AI labs such as Anthropic and OpenAI, competitor Cerebras and startups including Etched. Huang nevertheless maintained that Nvidia can serve every major model developer and therefore has broad visibility into demand.
For businesses planning AI capacity, the practical implication is to distinguish public enthusiasm from committed workloads, power availability and deployable systems. Nvidia’s forecast is based on its view of those inputs; procurement teams should apply the same discipline when assessing infrastructure commitments and supplier exposure.