Back

19/11/2026

NVIDIA doubles again: $115.8 billion and a $130 billion guide in the first quarter after Huang

NVIDIA revenue more than doubled to $115.8 billion as Vera Rubin began shipping. What the numbers and the CEO change mean for AI infrastructure plans.

On 17 November 2026, NVIDIA reported revenue of $115.8 billion for the third quarter of its fiscal year 2027, more than double the same quarter a year earlier. It also forecast $130 billion for the next quarter. The numbers say the AI infrastructure buildout is still accelerating. The context is new: this was the first report since Jensen Huang, who had led NVIDIA for 33 years, resigned as CEO the week before. For companies planning AI infrastructure for 2027, both halves of that story matter.

Key takeaways

  • Revenue was $115.8 billion (+103% year on year), and Data Center revenue was $107.6 billion (+110%), about 93% of the total.
  • NVIDIA guided Q4 to $130.0 billion, plus or minus 3%. That quarter has 14 weeks, and the forecast assumes no Data Center compute revenue from China.
  • Vera Rubin NVL72 has started shipping, and NVIDIA calls it its fastest product ramp ever.
  • Supply commitments rose to $390 billion, mainly for multi-year memory supply. Supply, not demand, is still the limiting factor.
  • CFO Colette Kress is now also interim CEO. The board is searching for a permanent CEO and has given no timeline.

The numbers: a beat and a $130 billion guide

NVIDIA now reports two segments: Data Center and Edge Computing. Data Center is split into Hyperscale and a second platform covering AI clouds, industrial and enterprise customers. The fastest growth came from that second group, which nearly tripled in a year.

Q3 fiscal 2027 (quarter ended 25 Oct 2026)ResultChange year on year
Total revenue$115.8 billion+103% (+20% on Q2)
Data Center$107.6 billion+110% (+21% on Q2)
– Hyperscale$60.1 billion+71%
– AI Clouds, Industrial & Enterprise$47.5 billion+195%
Edge Computing$8.2 billion+42%
Gross margin (GAAP / non-GAAP)74.8% / 74.9%Up on Blackwell Ultra mix
Diluted EPS (GAAP / non-GAAP)$2.70 / $2.68–
Q4 revenue outlook$130.0 billion ±3%14-week quarter

Two details deserve attention. First, the Q4 forecast includes an extra week, so part of the step from $115.8 billion to $130 billion is calendar rather than demand. Second, gross margin is expected to fall to 71–72% in Q4. NVIDIA says the slight dip from Q2 to Q3 reflected the start of the Rubin ramp, and new products usually carry lower margins early in their ramp.

NVIDIA also returned a record $28.1 billion to shareholders in the quarter through buybacks and dividends. The board increased the buyback authorisation by $150 billion in September, and $227.3 billion remained at quarter end.

Rubin ships: what Vera Rubin NVL72 means for the roadmap

Vera Rubin NVL72 is NVIDIA’s next rack-scale system after Blackwell. It combines Rubin GPUs and Vera CPUs in a single rack designed to work as one large computer. NVIDIA says production ramped and shipments began during the quarter, calling it “the fastest product ramp in the company’s history”. Coverage of the results names CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius among the providers already running Rubin racks.

The ramp is visible on the balance sheet too. Inventory rose from $31.6 billion to $39.7 billion “to support our Rubin product ramp”, and receivables rose partly because of shipment timing at launch. Blackwell Ultra is still selling, and NVIDIA is also selling Vera CPUs on their own.

For buyers, this means the first Rubin capacity will go to the largest clouds and AI cloud providers. Most enterprises will meet Rubin first as a cloud instance, not as hardware in their own data centre.

The quarter after Huang: Kress’s first report

NVIDIA announced on 12 November that Jensen Huang had resigned as President, CEO and board member, effective immediately. The company gave no reason. Colette Kress, Executive Vice President and CFO, became interim CEO while keeping her CFO role. Harvey C. Jones, a former CEO of Synopsys, became NVIDIA’s first-ever board chair. The board says it will consider internal and external candidates for a permanent CEO.

In the results release, Kress focused on continuity: “Our momentum is real,” she said, adding that “Vera Rubin is off to an explosive start, our roadmap is on track and our ambition has never been greater.” What the release and CFO commentary did not include is a timeline for the CEO search. NVIDIA’s forward-looking statements now list the leadership transition, the CEO search, and an ongoing investigation by an independent committee of the board as risks.

For customers, the near-term effect is limited. Products, contracts and the published roadmap continue. The longer-term question is strategic. Huang was closely identified with NVIDIA’s long-term bets on CUDA, networking and full-rack systems. A new CEO could change priorities on pricing, partnerships or how far NVIDIA goes into building and financing data centres itself.

Demand vs. supply: commitments, memory and capacity

The clearest sign that demand is still ahead of supply is how much NVIDIA is committing to buy. Supply commitments rose from $279 billion last quarter to $390 billion, “primarily related to the multi-year procurement of memory supply”. Including cloud service agreements, data centre leases and equity investments, total future commitments reached $518 billion.

NVIDIA is also helping customers secure physical capacity. It says land, power and data centre shells have become “the next critical phase” of the buildout, and that many AI cloud providers are growing faster than their balance sheets can support. It now has $32 billion in neocloud agreements and up to $118 billion in guarantees. Most of that ($105 billion) supports a large data centre campus in Ohio leased to OpenAI.

China remains a wildcard. NVIDIA shipped $2.4 billion of Hopper data centre products to China in Q3, but its Q4 forecast assumes no Data Center compute revenue from China at all. Any approved sales would come on top of the guide.

What it means for buyers and builders planning AI infrastructure in 2027

  • Plan around supply, not list prices. With memory locked up in multi-year contracts, GPU servers and even ordinary PCs are getting more expensive. NVIDIA notes that higher memory and system prices are already affecting consumer PC demand. Budget for longer lead times and higher prices for memory-heavy hardware.
  • Expect Rubin in the cloud first. If you need the newest generation in 2027, a reserved cloud capacity agreement is more realistic than buying racks. On-premises buyers should check what Blackwell or Blackwell Ultra systems their vendor can actually deliver.
  • Check power and cooling early. Rack-scale systems such as NVL72 need far more power and liquid cooling than standard servers. NVIDIA’s own spending on land, power and data centre shells shows where the bottleneck is moving.
  • Keep your software portable. Strong results and a leadership search together are a reminder not to depend on one vendor’s roadmap. Use containers, open-source frameworks and inference servers that can run on more than one type of accelerator where possible.
  • Watch the CEO search. A permanent appointment, and what the new CEO says about pricing, partners and data centre financing, will matter more for 2028 planning than any single quarter.

Planning GPU or AI infrastructure for 2027? Contact the LOGIQ team to talk through capacity, platform and migration options.

Sources

Photo: panumas nikhomkhai / Pexels