NVIDIA Q3 Fiscal 2027 Revenue Hits $115.8 Billion as Vera Rubin Ships Under Interim CEO

NVIDIA reported third-quarter fiscal 2027 revenue of $115.8 billion on Nov. 17, more than double the year-earlier figure, as data center sales reached $107.6 billion in the first quarter of volume shipments for its Vera Rubin platform. The Santa Clara, California, chipmaker released the results five days after Jensen Huang resigned as chief executive, making this the first earnings report under interim CEO Colette Kress, who also remains chief financial officer.
Key developments
- Revenue for the quarter ended Oct. 25, 2026, was $115.8 billion, up 20% from the prior quarter and 103% from a year earlier.
- Data center revenue was $107.6 billion, up 21% sequentially and 110% year over year. Edge computing revenue, which covers PCs, gaming, automotive and robotics, was $8.2 billion, up 42% year over year.
- GAAP gross margin was 74.8% and GAAP diluted earnings per share were $2.70, compared with $1.30 a year earlier. GAAP net income was $65.3 billion.
- Vera Rubin NVL72 racks began shipping and are running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius, according to the company.
- NVIDIA returned $28.1 billion to shareholders during the quarter through $22.1 billion of share repurchases and $6.0 billion of dividends.
Industry background
NVIDIA now reports two segments, Data Center and Edge Computing, rather than the separate gaming, professional visualization and automotive lines it used in earlier years. The new structure reflects how heavily the business leans on accelerated computing for cloud providers, AI model developers and so-called neoclouds. In the third quarter, data center sales made up roughly 93% of total revenue.
The quarter was the first in which Vera Rubin, the successor to the Grace Blackwell platform, shipped in volume. NVIDIA described the production ramp as the fastest in its history. Oracle Cloud Infrastructure made Vera Rubin NVL72 systems and Vera CPUs generally available, and Google Cloud launched Vera Rubin-based A5X instances. The company also introduced Rubin HPC, a GPU aimed at simulation and data processing, and DSX POD, a reference design for containerized AI factories that partners assemble and validate before delivery.
The results arrived during an unusual leadership period. On Nov. 12, NVIDIA disclosed that Huang had resigned as president, CEO and director with immediate effect, that Kress would serve as interim CEO and that director Harvey C. Jones would become the board's first chair. The company said it would search for a permanent CEO among internal and external candidates. In the earnings release, Kress said the company's momentum was real and that Vera Rubin was off to an explosive start, with its roadmap on track.
Industry implications
For cloud operators and AI developers, the figures confirm that spending on AI training and inference capacity continued to rise through the fall. Free cash flow of $43.8 billion in a single quarter gives NVIDIA wide room to fund supplier prepayments, buybacks and acquisitions at the same time. The company listed its agreement to acquire Hugging Face, the open model and dataset platform, among the quarter's highlights. The deal would extend its reach from hardware and system software into the developer community that distributes open models.
The customer list in the release also signals how demand is spreading. NVIDIA named Meta, OpenAI, Nscale, Verda and others as builders on its Vera CPU, and said Meta plans to deploy Vera Rubin, Vera CPUs, Spectrum-X networking and BlueField-4. Anthropic expanded access to Grace Blackwell and Vera Rubin capacity through cloud partners. In Australia, the company announced a collaboration on infrastructure for a planned buildout of up to 2 gigawatts of DSX AI factories by 2027.
Competitive and supply chain implications
On the same day as its results, NVIDIA expanded commitments with Amkor Technology, Foxconn, Kinsus Interconnect Technology, Samsung Electro-Mechanics, Siliconware Precision Industries and TSMC. Kinsus agreed to expand ABF substrate capacity dedicated to NVIDIA by more than 3.5 times at its Youth campus in Taoyuan, Taiwan, with production starting in the second half of 2027. SPIL signed a multiyear agreement, backed by multibillion-dollar NVIDIA funding, to increase dedicated advanced packaging capacity in Taiwan by mid-2029. Multiyear agreements with Amkor announced in July and October total nearly $3 billion, including prepayments for capacity in Arizona and South Korea.
These deals follow October commitments covering substrates, testing and optical engines with IBIDEN, King Yuan Electronics, Unimicron and others. The pattern is clear: substrates, CoWoS-class packaging and testing remain the constraints, and NVIDIA is using its balance sheet to lock in capacity ahead of rivals. Competing accelerator designers and custom chip programs at hyperscalers draw on many of the same suppliers, so dedicated capacity agreements can tighten availability for other buyers.
NVIDIA is also widening its platform to partners. MediaTek deepened its partnership to extend NVIDIA's rack-scale platform to custom XPUs through NVLink Fusion, and Cisco announced USFP, a new optical module form factor for AI factory networks.
Regional implications
In Asia, the supply chain announcements concentrate new investment in Taiwan and South Korea, where Samsung Electro-Mechanics amended its substrate agreement covering factories in Sejong and Busan. Amkor's prepayments support capacity in both Arizona and South Korea, giving the United States a larger role in advanced packaging. In Europe, NVIDIA cited adoption of its DSX designs to cut AI factory power demand and cooling water use, and it ran a record simulation of a fault-tolerant quantum processor on the JUPITER supercomputer at Germany's Jülich Supercomputing Centre.
Outside the data center, the automotive work highlighted in the quarter is weighted toward Europe. NVIDIA expanded automated driving partnerships with Mercedes-Benz and Volvo Cars, and said Bolt with Lucid, Volkswagen Commercial Vehicles and Einride are showcasing its Hyperion platform for automated vehicles and robotaxi fleets.
Credence Research analysis
The quarter shows that NVIDIA's operating machine is running independently of the leadership change, at least for now. Vera Rubin moved into volume shipments, gross margin held near 75% through a platform transition, and the company kept returning cash at scale. The real test of the interim structure will come over the next two to three quarters, as customers and suppliers sign multiyear commitments that typically rest on confidence in a long-term product roadmap and a stable executive team.
The supply chain announcements may matter more than the headline revenue. By funding substrate and packaging expansions that come online between 2027 and 2029, NVIDIA is converting cash flow into reserved capacity at the narrowest points of the AI hardware chain. That strategy raises the cost of entry for competitors that depend on the same Taiwanese and Korean suppliers, but it also commits NVIDIA to large prepayments that assume demand stays strong across the next platform cycle.
What comes next
- NVIDIA will pay its next quarterly dividend of $0.25 per share on Dec. 23, 2026, to shareholders of record on Dec. 2.
- The board's search for a permanent CEO is underway, with internal and external candidates under consideration.
- The fourth quarter of fiscal 2027 includes an extra week, which will affect year-over-year comparisons.
- Kinsus capacity begins production in the second half of 2027, and SPIL's expanded packaging capacity is due by mid-2029.
- Closing of the Hugging Face acquisition and further Vera Rubin deployments, including at SpaceXSI, are the next milestones to watch.
Sources
- NVIDIA: NVIDIA Announces Financial Results for Third Quarter Fiscal 2027
- NVIDIA: NVIDIA Expands Commitments With Supply Chain Partners to Support AI Infrastructure Buildout
Image: NVIDIA's Endeavor building in Santa Clara, California, by Crocodiletiger via Wikimedia Commons, CC BY-SA 4.0. Cropped and edited by Credence Research.
About the Author

Senior News Editor at Credence Research Inc