Jensen Huang set the number himself. In remarks reported by Bloomberg on September 17, the Nvidia chief executive said the company’s chip sales in 2027 will reach twice the level of 2026. The statement landed before the market opened, and Nvidia shares rose more than 2 percent in premarket trading.
The claim arrived at a moment when the market had begun to doubt the AI boom. In the days before, investors had sold off memory and semiconductor-equipment stocks on signs that spending on artificial intelligence might be cooling. Huang’s figure was a direct answer to that doubt, delivered in the company’s favored currency: an enormous number.
Huang has been raising the bar steadily. He previously lifted the company’s 2027 sales target into the trillion-dollar range, a figure that would have seemed fanciful a few years earlier and now functions as the base case for the most valuable company in the world. The doubling claim extends that logic one more step.
The math behind the confidence is straightforward in Huang’s telling. Every generation of Nvidia’s chips is faster than the last, and each gains enough in efficiency that customers can justify replacing the previous generation sooner. The company’s own roadmap, he argues, makes the demand curve almost arithmetic.
The counterargument is equally simple and has been getting louder. Some analysts and investors warn that the spending is financed by debt and hope, that the large models may not yet earn back what they cost to run, and that a pullback in any one customer could ripple through the whole supply chain.
Huang has heard this argument before and rejected it each time. His position, repeated across earnings calls and conferences, is that the world is only at the beginning of a shift from general-purpose computing to accelerated computing, and that the transition will take years, not quarters.
The trillion-dollar ambition has concrete components. Nvidia’s data-center business, its largest and fastest-growing line, is driven by the hyperscalers buying GPUs by the tens of thousands, and by a new class of AI-native startups renting compute rather than building their own.
Barron’s ran the same day with a headline that read “AI Demand Has Not Slowed,” a formulation that captured the mood the company was trying to set. The message, coordinated or not, was that the sellers in the previous sessions had misread the trajectory.
The stakes extend far beyond Nvidia. The company has become the reference point for the entire AI trade, and its guidance moves the valuations of memory makers, networking suppliers, and the utilities that power the data centers. A claim about doubling sales is, in effect, a claim about the whole sector.
Huang’s numbers have a history of holding up. The company has repeatedly exceeded its own forecasts as demand has accelerated, and its ability to convert orders into revenue has become one of the more reliable constants in the technology industry.
The risk is that the company’s scale now works against it. Doubling from a base of tens of billions of dollars in chip sales requires a volume of orders that few buyers can place, and the concentration of those buyers among a handful of hyperscalers makes the revenue more fragile than the headline number suggests.
Analysts noted that Huang’s statement is a projection, not a commitment, and that the gap between a chief executive’s optimism and the order book can close suddenly. The company has not changed formal guidance, and the remarks carried the informal authority of the man who runs it.
The supply chain has its own limits. Leading-edge chip manufacturing is constrained by the foundries’ capacity and by the advanced packaging that binds memory to logic. Even if demand doubles, supply must be able to follow, and that depends on factories being built years in advance.
Huang’s broader argument is that the constraints are being solved as fast as they can be solved. Nvidia has spent the year diversifying its suppliers and its customers, and the company frames the doubling target as a function of capacity it has already secured.
The market’s reaction was swift but contained, a sign of how much of the future is already priced in. A 2 percent premarket gain is small for a company whose shares have multiplied in recent years, and it suggests investors are waiting for the next earnings report to confirm the talk.
For the doubters who sold memory stocks days earlier, Huang’s message was unmistakable: the slowdown they feared is not in his forecast. Whether the forecast holds is a question that will be answered by the company’s own results, quarter by quarter, through 2027.
The broader economy now watches these projections the way it once watched oil. Nvidia’s numbers have become a leading indicator for capital spending, electricity demand, and the health of a bull market that has leaned on a single story. When Huang talks about doubling, the audience is far wider than his shareholders.
What Huang did not say is as significant as what he did. He offered no new product, no new customer, no revision to the current quarter. The statement was a forecast about scale, and scale is the one variable Nvidia has proven it can compound.
