Analyst(s): Futurum Research
Publication Date: November 25, 2025
NVIDIA’s quarter underscores continued AI infrastructure demand across hyperscalers, model builders, and enterprises. Management highlighted Blackwell momentum, networking scale, and deepening ecosystem partnerships as core drivers into FY 2026.
What is Covered in this Article:
- NVIDIA’s Q3 FY 2026 financial results
- Blackwell ramp and CUDA adoption
- Networking scale and NVLink Fusion
- Ecosystem and customer commitments expand
- Guidance and Final Thoughts
The News: NVIDIA (NASDAQ: NVDA) announced financial results for Q3 FY 2026. Revenue was $57.0 billion, up 62% year over year (YoY), versus Wall Street consensus of $55.2 billion (beat by 3.3%). Data Center revenue came at a Q3 record of $51.2 billion, up 66% YoY; Gaming was $4.3 billion, up 30% YoY; Professional Visualization was $760 million, up 56% YoY; Automotive was $592 million, up 32% YoY; and OEM & Other was $174 million, up 79% YoY. Non-GAAP gross margin was 73.6%, down 1.4 points YoY. Non-GAAP operating income was $37.8 billion, up 62% YoY. Non-GAAP net income was $31.8 billion, up 59% YoY, and non-GAAP diluted EPS was $1.30, up 60% YoY.
“Blackwell sales are off the charts, and cloud GPUs are sold out,” said Jensen Huang, founder and CEO of NVIDIA. “Compute demand keeps accelerating and compounding across training and inference — each growing exponentially. We’ve entered the virtuous cycle of AI.”
NVIDIA Q3 FY 2026: Record Data Center Revenue, Higher Q4 Guide
Analyst Take: NVIDIA’s Q3 FY 2026 results reflect durable, multi-segment AI infrastructure demand anchored by the Blackwell platform, CUDA software stack, and a growing network and systems footprint. The mix continues to skew toward Data Center, with networking scale and NVLink-led architectures expanding attach across hyperscalers and sovereign builds. A widening set of ecosystem alliances and strategic investments signals deeper workload optimization and longer-term visibility. With supply plans aligned to an annual cadence and the Rubin platform on track for H2 FY 2027 ramp, the near-term setup supports elevated utilization and mid-70s margin targets.
Blackwell Ramp, CUDA Moat, and Workload Shift
Blackwell GB300 shipments have crossed GB200 and now account for roughly two-thirds of Blackwell revenue, with installed base utilization spanning Blackwell, Hopper, and Ampere. Management emphasized that accelerated computing is replacing CPU-only approaches across data processing, search, recommendations, and content understanding as generative and agentic AI workloads expand. CUDA compatibility and continual software optimization extend useful life, with A100 systems still fully utilized six years post-ship. On benchmarks, Blackwell Ultra delivered 5x faster time-to-train versus Hopper on MLPerf, and on DeepSeek-R1 mixture-of-experts, Blackwell achieved 10x performance per watt and 10x lower cost per token versus H200. NVIDIA Dynamo adoption by major clouds underscores software-led throughput gains for complex inference at scale. Together, these dynamics support continued platform preference and spend consolidation on NVIDIA.
Customer Commitments and Ecosystem Expansion
NVIDIA outlined multi-gigawatt commitments that broaden beyond hyperscalers to model builders, sovereigns, and enterprises. The company is working with OpenAI to deploy at least 10 gigawatts of NVIDIA systems and with Anthropic to adopt an initial 1 gigawatt of Grace Blackwell and Vera Rubin systems. The quarter featured announcements spanning xAI’s Colossus 2 gigawatt-scale data center, AWS/HUMAIN deployments including up to 150,000 accelerators, and an aggregate of roughly 5 million GPUs across AI factory projects. Enterprise platforms such as ServiceNow, SAP, and Palantir (with CUDA-X and AI models) are integrating NVIDIA’s stack, while industry use cases in healthcare and productivity (such as Abridge, OpenEvidence, and Cursor) point to ROI-led adoption. Management cited OpenAI weekly users at 800 million and Anthropic ARR at $7 billion as indicators of end-market traction. These commitments suggest durable demand signals that extend visibility well into calendar 2026.
Networking and System-Scale Differentiation
Networking revenue reached $8.2 billion, up 162% YoY, with NVLink scale-up and strong growth across Spectrum-X Ethernet and Quantum-X InfiniBand. Management noted Ethernet GPU attach roughly on par with InfiniBand, with Meta, Microsoft, Oracle, and xAI building gigascale AI factories on Spectrum-X. The new Spectrum-XGS “scale-across” capability enables giga-scale factories, complementing scale-up and scale-out options unique to NVIDIA’s portfolio. NVLink Fusion partnerships with Fujitsu and Intel, along with Arm integrating NVLink IP, expand cross-ecosystem CPU-GPU coherence options. BlueField-4 launched as the data center “operating system” processor for AI factories, and Rubin remains on track to ramp in the second half of FY 2027. This end-to-end architecture further differentiates NVIDIA on performance, TCO, and time-to-deploy at rack and factory scale.
Guidance and Final Thoughts
For Q4 FY 2026, NVIDIA guided revenue to $65.0 billion (plus or minus 2%; estimate: $62 billion), non-GAAP gross margin of 75.0% (mid-70s targeted into FY 2027; estimate 74.6%), non-GAAP operating expenses of $5.0 billion (estimate: $4.6 billion), and a 17% tax rate. Management reiterated zero data center compute contribution from China in the outlook while highlighting continued Blackwell momentum and ongoing supply readiness. Partnerships with OpenAI and Anthropic reinforce a strategy of co-optimization across models, systems, and software to sustain throughput and efficiency gains. With announced AI factory projects approximating 5 million GPUs and networking scaling across NVLink, Ethernet, and InfiniBand, NVIDIA’s full-stack approach strengthens visibility across cloud, sovereign, and enterprise builds. The upcoming Rubin platform and continued annual cadence provide a path to performance-per-dollar improvements and broader workload coverage.
See the full press release on NVIDIA’s Q3 FY 2026 financial results on the company website.
Declaration of generative AI and AI-assisted technologies in the writing process: This content has been generated with the support of artificial intelligence technologies. Due to the fast pace of content creation and the continuous evolution of data and information, The Futurum Group and its analysts strive to ensure the accuracy and factual integrity of the information presented. However, the opinions and interpretations expressed in this content reflect those of the individual author/analyst. The Futurum Group makes no guarantees regarding the completeness, accuracy, or reliability of any information contained herein. Readers are encouraged to verify facts independently and consult relevant sources for further clarification.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other insights from Futurum:
NVIDIA Q2 FY 2026 Earnings: Networking Steals the Spotlight and Q3 Ramp Will Be Key To Watch
Ansys and Synopsys Expand Simulation Capabilities with NVIDIA Omniverse Integration
Will NVIDIA and AMD Pass the 15% Fee to Chinese Buyers?
Author Information
Futurum Research
Futurum Research delivers forward-thinking insights on technology, business, and innovation. Content published under the Futurum Research byline incorporates both human and AI-generated information, always with editorial oversight and review from the expert Futurum Research team to ensure quality, accuracy, and relevance. All content, analysis, and opinion are based on sources and information deemed to be reliable at the time of publication.
The Futurum Group is not liable for any errors, omissions, biases, or inadequacies in the information contained herein or for any interpretations thereof. The reader is solely responsible for any decisions made or actions taken based on the information presented in this publication.
