NVIDIA's Jensen Huang Just Said Something the Market Hasn't Priced In — Or Has It?

"I expect Nvidia to sell twice as many chips next year as we do this year," Jensen Huang told reporters Thursday on the sidelines of an AI summit in Scotland.

That's the number that landed first. Not a revenue figure, not a margin projection, not guidance measured in the currency investors actually use to value companies. Volume. Twice as many chips.

Huang has never been shy about projecting demand. But he also has never publicly disclosed the total number of chips NVIDIA ships. So this was a new benchmark drawn in a metric the company normally keeps internal. And it arrives two weeks after NVIDIA already stunned analysts with its first-ever year-ahead revenue forecast: 70% growth for fiscal 2028, implying roughly $673 billion — well above the $570 billion consensus had modeled.

Here's the thing the market is wrestling with: 70% is what management committed to. "Twice as many" is what they believe demand justifies. The gap between them is supply.

At the August earnings call, Huang was explicit. Demand is "much greater than 70%," with customer forecasts, per CFO Colette Kress, pointing to growth roughly doubling. The 70% figure was chosen because it's the number the supply chain can deliver — constrained largely by memory components. "Our entire supply chain is challenged, and it's everybody," Huang said. "Everybody is really running flat out." Supply and capacity commitments have surged from $119 billion to $279 billion, driven by rising memory costs.

Put differently: the 70% guide is the floor demand can force the company to hit, not the ceiling.

NVIDIA stock (NASDAQ: NVDA) is trading at $222.27 today, up 1.336%, after jumping 2.8% Thursday on the Scotland comments. The market cap sits at $5.36 trillion. The stock has gained 19.2% year-to-date and 32.7% over the past four months.

That $5.36 trillion figure deserves a moment. NVIDIA trades at a trailing P/E of 27.8 — which looks reasonable until you realize the forward P/E is 59.3. The market is already pricing in an extraordinary earnings acceleration. Growth isn't assumed at this valuation. It's required.

The arithmetic behind the current price works only if the company sustains roughly what it just guided. The trailing P/E of 27.8 reflects what NVIDIA has already delivered: gross margins of 74.15%, operating margins of 64.02%, return on invested capital of 86.65%, and free cash flow of $127 billion over the trailing twelve months. On those metrics, the business is operating at a level that doesn't show up very often in corporate America.

But the forward P/E of 59.3 says investors believe those margins will hold — or at least not deteriorate enough to hurt — while revenue climbs 70% next fiscal year. NVIDIA itself guided gross margins to bottom at 71-72% in the fourth quarter, then settle at 72-73% in fiscal 2028. That's a step down from the current 74.15%, but management says incoming price increases will cushion the decline. AMD, for its part, just notified partners of a 10% price increase starting in the fourth quarter to offset rising TSMC wafer costs — suggesting the entire semiconductor supply chain is pricing capacity at a premium.

Now, back to the "twice as many" number.

The math of volume versus revenue matters here. If NVIDIA doubles the number of chips it ships but the product mix shifts toward lower-average-selling-price components — CPUs, switch chips, laptop processors, the Nintendo Switch 2 chip — revenue growth could fall well short of 100%. Conversely, if the Vera Rubin platform, which NVIDIA began shipping in early August and which targets higher-margin "agentic AI" workloads, takes up a larger share of volume, revenue could accelerate faster than unit counts suggest.

We don't know the mix. NVIDIA doesn't disclose it. That's part of why Huang chose volume as his metric. It's the one number he can state with confidence independent of product pricing decisions that haven't been made yet.

There's also the question of what "next year" means. Huang didn't specify fiscal year versus calendar year, and his fiscal year ends in January. If "next year" runs from October 2026 through September 2027, that window captures only about three quarters of fiscal 2028 — meaning the second half of fiscal 2028 would have to carry enormous weight for the doubling to work. We aren't really sure.

The timing adds another layer. The Federal Reserve raised interest rates by 25 basis points on Wednesday — the first hike in over three years, to a range of 3.75%-4% — citing inflation driven by oil prices, tariffs, and, yes, increased AI spending. Sixteen of 18 Fed participants expect another hike before year-end, and the 10-year Treasury yield has climbed roughly 100 basis points from its February low.

Normally, tighter monetary policy weighs on high-multiple growth stocks. NVIDIA's rally through this cycle suggests something else is going on: the growth story is outrunning the rate story, at least for now.

What about the risk case? The competition angle hasn't gone away. Amazon, Alphabet, Microsoft, and Meta are all developing custom AI chips. If major customers migrate even a fraction of workloads to cheaper, specialized alternatives, NVIDIA faces margin compression even if total volume holds. And the 70% guidance explicitly excludes China data center revenue — a market that could be a tailwind or a wild card depending on how export restrictions evolve.

Still, the earnings scorecard is hard to argue with. Over the past eight reported quarters, NVIDIA has beaten EPS estimates every single time and revenue estimates every single time. The most recent quarter — Q2 fiscal 2026 — came in at $46.7 billion in revenue (versus $46.0 billion consensus) and $1.05 in EPS (versus $1.01 expected). The non-hyperscaler segment, which now represents roughly half the business, is growing at 100% annually.

To put a bow on the arithmetic: at 59 times forward earnings, NVIDIA needs to deliver on the order of 70% revenue growth while holding margins above 71%. Huang is signaling demand could push well past that — but whether the supply chain can catch up, and whether the product mix at doubled volume sustains today's margins, is the live question.

The market seems to believe the first part. Whether it's pricing in the second part is harder to tell.