Nvidia forecasts $673B in FY2028 sales

Nvidia reported $96.2 billion in second-quarter fiscal 2027 revenue on August 26, roughly double the same quarter a year earlier, with data-center revenue up 117% to $89 billion. On the earnings call, CFO Colette Kress gave the company’s first-ever forecast extending into fiscal 2028, projecting 70% revenue growth for that year, which implies roughly $673 billion in annual sales if Wall Street’s current fiscal 2027 consensus holds. That figure dwarfs the 44% average growth analysts had been modeling, according to LSEG data. CEO Jensen Huang told investors:

Our demand is much greater than 70%. Our supply allows us to confidently deliver 70%, and we’re going to continue to work with our supply chain to increase on that.

— Jensen Huang, Nvidia CEO

The guidance is notable less for its size than for what it says out loud: Nvidia is framing its own manufacturing and supply-chain capacity, not customer demand, as the binding constraint on how fast the AI buildout can grow. Huang’s phrasing implies Nvidia believes it could sell more than $673 billion worth of chips next fiscal year if it could make them, and that the number Wall Street should model is a supply ceiling rather than a demand estimate.

What it means for you

If your roadmap assumes GPU access or inference pricing gets meaningfully easier over the next 12-18 months, Nvidia’s own management just told you the constraint is supply, not appetite. That matters for anyone signing multi-year compute commitments right now, since the labs and clouds racing to lock in capacity, including Anthropic’s own just-signed $45 billion deal with Nscale, are competing against a company that says it is selling everything it can build. Budget compute costs on the assumption that GPU scarcity, not GPU price competition, is the more likely story through 2027.