Jensen Huang put a number on his company’s trajectory. The Nvidia chief executive said on September 17 that the company expects to sell roughly twice as many chips next year as it will this year, a forecast that extends the extraordinary run of demand for the processors that power artificial intelligence.
The statement was a projection of volume rather than revenue, and Huang offered it without the usual hedging. Nvidia’s chips are the core component of the AI boom, and a doubling of unit sales would imply a continuation of the industry’s buildout at a pace that has already made Nvidia one of the world’s most valuable companies.
The comment came alongside a second point Huang has been making with increasing emphasis. AI safety, he said, is critical, and developers should test their AI systems rigorously, delaying the release of products that are not safe enough. The pairing of a growth forecast with a call for caution captures the two sides of Huang’s public role: the salesman of the AI buildout and the elder statesman warning about its risks.
Huang has emerged as a prominent voice in the safety debate that has divided the industry’s leadership. His position has been that new laws are not the answer, and that the companies building AI can be trusted to test their own systems before release. The September 17 remarks restated that view in the plainest terms, urging developers to hold back products that do not meet a safety bar.
The forecast of doubled chip sales aligns with what Nvidia’s customers have been signaling. The largest technology companies have announced data center spending plans that run into the hundreds of billions of dollars, and the bulk of that capital flows, directly or indirectly, into the kind of processors Nvidia makes. If those spending plans hold, Nvidia’s unit sales would have to grow sharply just to meet the demand that is already contracted.
Supply, not demand, has been the constraint on Nvidia’s growth for most of the past two years. The company’s chips have been sold out far in advance, and customers have complained about waiting for delivery. A doubling of volume would require Nvidia to have solved enough of its supply chain bottlenecks to actually ship that many more units, a question the company has been working to answer through expanded manufacturing capacity.
The forecast also carries a message about the durability of the AI buildout. Doubling chip sales implies that Huang does not see the current spending as a bubble that will deflate next year, but as a trend with at least another year to run. That is a bet on the willingness of Nvidia’s largest customers to keep spending at current levels or beyond.
The chip volumes Huang described cover Nvidia’s full product range, from the data center accelerators that dominate its revenue to the graphics and networking products that round out the portfolio. The AI accelerators are the engine of growth, and their share of the total is expected to keep rising as the data center business expands.
Huang did not provide a specific baseline for the “twice as many” forecast, which leaves room for interpretation about the exact denominator. But the direction is unambiguous, and it matches the message Nvidia has delivered in recent quarters: the demand for its chips is growing faster than the company can currently supply them, and the path forward is a steep one.
The data center business is now the engine of Nvidia’s growth. The company’s accelerators have become the standard component for AI computing, and the segment’s revenue has grown to dominate the company’s results. A doubling of unit sales next year, if it comes to pass, would extend a run that has already made Nvidia the most valuable company in the world at points during the past year.
The supply chain is the question mark behind the forecast. Nvidia’s chips have been sold out well in advance for most of the past two years, and the company has worked to expand manufacturing capacity with its supplier, TSMC, and its partners. Doubling the number of chips sold means doubling what the supply chain can deliver, a constraint that is not entirely within Nvidia’s control.
Huang’s safety remarks have become a signature of his public appearances. He has argued that the companies building AI are capable of testing their own systems and that new laws are not the answer, a position that places him on one side of an increasingly public disagreement among the industry’s leaders. The forecast and the caution, delivered together, are the two halves of the case he has been making for years: the technology is coming fast, and it can be handled.
The remarks were made at an event where Huang also touched on the safety theme that has become a fixture of his appearances. The combined message, growth and caution delivered in the same breath, is the posture Nvidia has adopted as the company that both powers and worries about the AI era.
