Key Takeaways
- Anthropic’s S-1, filed Monday, shows $2.08 of revenue per dollar of compute, up from $0.40 a year earlier, according to an analysis Gavin Baker endorsed.
- Baker argues that cheaper tokens raise total spending on compute, which supports demand and pricing for NVIDIA’s GPUs.
- Analysts expect NVIDIA’s revenue to grow from $215.9 billion in fiscal 2026 to $692 billion in fiscal 2028, but rising memory costs are expected to cut gross margin to 71% to 72% in the January 2027 quarter.
- NVIDIA trades at 18x forward earnings, compared with a three-year average of 30x, ahead of its Nov. 17 report.
Anthropic’s S-1 came out Monday. It gave NVIDIA (NVDA) bulls the number they’d been waiting for.
Gavin Baker, who runs Atreides Management, had said the filing would “break a lot of people’s brains,” because so many macro and value investors assume “tokens are subsidized.” On Tuesday, he posted: “Brains were indeed broken.”
The day before, he’d made the case on X:
“I am shocked, absolutely shocked, that Anthropic ARR has likely gone vertical in October after they cut token prices. Except the exact same thing happened for the entire AI market, which is why GPU rental prices have surged while token prices declined.”
That’s a big claim. It’s also the whole bull case for NVIDIA in two sentences.
TIKR tracks 62 analysts on Nvidia. How many say “Buy”?
See the full breakdown of Buy, Hold, and Sell ratings on Nvidia, along with Wall Street’s revenue and earnings estimates, all in one place. More than 500,000 investors use TIKR to research stocks like this one. See Nvidia’s Analyst Ratings
The filing backs him up
The October number is Baker’s own estimate, since the S-1’s figures stop at June 30. But what the filing does show points the same way.
An X account, @AGTPinsights, pointed out that the filing shows “$2.08 of revenue per compute dollar now versus $0.40 a year ago, even with lower prices.” Baker answered with one word: “Indeed.”
So each dollar Anthropic spends on compute now brings in about five times as much revenue as it did a year ago, even though prices are lower. And the company is profitable: GAAP operating income was $974 million in the second quarter, an 8% operating margin.
Baker says every data point on AI so far has shown this kind of elasticity: lower prices lead to so much more use that total spending goes up. He also argues that revenue per GPU can rise as token prices fall, as long as each GPU produces enough extra tokens. If what Anthropic pays for GPUs rises by less than that combined effect, both its revenue and its margins can grow.
Good news for whoever sells the GPUs
If cheaper tokens keep pushing up what labs spend on compute, the chipmaker wins. You can already see it in rents: Social Capital puts H100 rental prices up 45% since December.
Anthropic matters directly here, too. NVIDIA invested $10 billion in the lab, and management said in August that it had almost no business with Anthropic until recently. Anthropic now runs on NVIDIA systems at scale.
Wall Street already expects NVIDIA’s revenue to keep climbing over the next few years.
Analysts expect revenue of $692 billion in fiscal 2028, up from $215.9 billion in the year ended January 2026. Management calls its own guide for that year, roughly 70% growth, a supply-limited figure. If Baker is right about elasticity, how many chips NVIDIA can make sets the limit on those estimates, and nothing in Anthropic’s filing argues for cutting them.
Pricing power shows up in the margin. NVIDIA’s gross margin has stayed above 70% for three straight fiscal years while token prices kept falling.
The pressure ahead comes from memory costs. Management expects gross margin to bottom at 71% to 72% in the January 2027 quarter as memory prices climb. It then sees 72% to 73% in fiscal 2028, once its own price increases take effect.
The stock is priced for a peak
One lab’s elasticity can’t tell you whether the industry builds more GPUs than it can use. Citadel’s Ken Griffin has warned that if supply ever outruns demand, “marginal pricing will actually go into freefall.”
The stock already reflects plenty of that worry. NVIDIA trades at 18x forward earnings, just above the 17x low it hit on Oct. 23 and far below its three-year average of 30x.
My take: Anthropic’s filing is the best evidence so far that cheaper AI leads to more spending on compute, and that’s what keeps GPU rents rising. At 18x, NVIDIA is priced as if that spending is close to its peak, and I’d side with Baker.
Of course, the October surge is still Baker’s estimate. NVIDIA’s Nov. 17 report is the next real test, including its margin guide for the January quarter.
So what is NVIDIA stock actually worth?
TIKR lets you forecast the future price of any stock in less than a minute. Just enter a few assumptions into TIKR’s valuation model and see what NVIDIA could be worth. Start from Wall Street consensus estimates, or adjust the inputs to reflect your own view of the business. It’s free to use.
Disclaimer:
Please note that the articles on TIKR are not intended to serve as investment or financial advice from TIKR or our content team, nor are they recommendations to buy or sell any stocks. We create our content based on TIKR Terminal’s investment data and analysts’ estimates. Our analysis might not include recent company news or important updates. TIKR has no position in any of the stocks mentioned. Thank you for reading, and happy investing!
