Nvidia’s October results were another example of a company accelerating growth in the face of the law of large numbers. Revenue of $57B was up 62% y/y, beating the whisper by 2.7% and the in-print estimates by 3.8%. Data Center revenue was up 66% y/y and the company delivered a record $10B sequential total revenue increase, or 22% y/y.

On guidance, Nvidia raised its January quarter outlook to $65B in revenue,  implying roughly 65% y/y growth, up from 62% in October. Management also noted that the ramp of the newer GB300 part is already contributing roughly two thirds of total Blackwell revenue, and they have more demand than they can supply.

Below the surface, the October quarter and January guide remind us that the collective “brain” of AI, defined as GPUs in use, is going to be bigger than we thought three months ago. The bigger that AI brain becomes, the more use cases appear, which in turn strengthens and extends the AI investment cycle. Put differently, this beat and raise suggests the AI trade has more room to run than skeptics of capex sustainability believe.

The most important part of the call for the longer term story was management’s reiteration of comments from Oct 28th that they have  visibility into $500B of Blackwell and Rubin revenue through the end of calendar 2026, and that excludes any additional bookings and excludes any revenue from China.

CFO Colette Kress reiterated that they are “on track” for that forecast and emphasized that “the number will grow” as new deals are signed. In practical terms, that $500B pipeline implies at least 54% y/y revenue growth in calendar 2026, compared to the Street’s revised higher number that now calls for 48% growth.

Colette also underscored that the three scaling laws of AI pre training, post training, and inference are intact and are feeding a positive virtuous cycle: more compute drives better intelligence, which drives greater adoption and profits, which funds more compute. Jensen added color here, noting that pre training and post training continue to scale and that chain of thought and reasoning are making inference far more compute intensive than many investors appreciate.

The topic of physical AI came up as well. Colette described physical AI as already a multibillion dollar business and “the next leg of growth” for Nvidia, pointing to robotics and factory automation as examples. This is an important point. Wall Street is largely focused on how long the current generative AI buildout will last, but the physical AI buildout is a separate wave that can kick in over the next several years and help sustain the company’s growth above consensus for longer.

Taken together, the $500B revenue pipeline, the intact scaling laws, and the emergence of physical AI support a view that Nvidia can grow revenue at least 54% next year with upside, ahead of where the Street is today.

China was again a point of focus, but the practical takeaway is that it doesn’t matter for the current guidance. In the October quarter, Nvidia’s H20 was immaterial and management noted that “sizable purchase orders never materialized” due to geopolitical issues and an increasingly competitive market in China.

Looking ahead, Colette was explicit that January guidance and the $500B revenue expectations for Blackwell and Rubin assumes no China revenue, consistent with prior quarters.

Nvidia has in the past emphasized that it remains committed to engagement with both the US and Chinese governments and views China as an important market for the long term, and it could account for 20% of overall revenue someday.

The question whether Nvidia can grow revenue at a 50% plus rate while maintaining profitability remains in the face of higher input costs for advanced nodes and memory is a debate. The company reported October gross margin of  73.6%, slightly above its guidance. For January, the company guided to 75%  gross margin, plus or minus 50 basis points.

More important than the quarter to quarter beats is the outlook for next year. Colette said they are working to hold gross margins in the “mid 70s” in fiscal 2027, even as input prices rise and systems become more complex. She pointed to ongoing efforts in cost improvement including product mix as levers (seller more high ASP GPUs)  to offset higher component costs. In the Q&A, she also reminded investors that Nvidia plans its supply chain and negotiates with partners well in advance, and that long standing relationships and scale give the company leverage when locking in pricing and capacity.

If Nvidia can keep gross margins roughly where they are today while revenue grows at a rate north of 50%, the earnings power of the model is higher than the Street is currently expecting.

On one side, investors worry about overbuilding and overspending on AI infrastructure. On the other, Nvidia continues to report that demand is ahead of supply. Colette said demand for AI infrastructure “continues to exceed our expectations,” given the cloud providers are fully utilized and ordering more GPU’s than the company can supply.

To prepare for what they describe as “significant growth ahead,” Nvidia grew inventory 32% q/q and increased supply commitments 63% sequentially. Jensen noted that the company has planned its supply chain alongside partners like TSMC, memory vendors, and system ODMs for a “big year”.

On the demand side, the Kress highlighted that the mega cap tech company’s, which I estimate account for 55% of sales, are increasingly adopting Nvidia networking solutions.

When asked directly about concerns that AI capex could be overbuilt, Jensen’s message was effectively “don’t overthink it.” He pointed out that AI applications are among the fastest growing in the world, that the number of companies building on them continues to rise, and that pre training, post training, and inference needs are still in the early stages. Hundreds of billions in capex, in his view, will quickly become profitable as AI drives both cost reductions and new revenue streams. My take: I think he’s right.

In the near term, it’s clear Nvidia can’t keep up with demand. The company did not give guidance or any color on when they’ll reach supply–demand equilibrium. My sense is it will be sometime in CY27. From an NVDA stock perspective, beating numbers and raising guidance while being sold out is an ideal position because it gives investors room to dream about how strong the business could be if they were able to make enough GPUs.

Disclaimer