We delivered another outstanding quarter with revenue of $57 billion, up 62% year-over-year, and a record sequential revenue growth of $10 billion, or 22%. Our customers continue to lean into three platform shifts, fueling exponential growth for Accelerated Computing, powerful AI models, and Agentic applications. We currently have visibility to $500 billion in Blackwell and Rubin revenue from the start of this year through the end of Calendar Year 2026. Record Q3 Data Center revenue of $51 billion increased 66% year-over-year, a significant feat at our scale.

Compute grew 56% year-over-year, driven primarily by the GB300 ramp, while networking more than doubled given the onset of NVLink scale-up and robust Double-digit growth across Spectrum-X Ethernet and Quantum-X InfiniBand. While Anthropic recently reported that its annualized run rate revenue has reached $7 billion as of last month, up from $1 billion at the start of the year. Companies such as Cursor, Anthropic, OpenEvidence, Epic, and Abridge are experiencing a surge in user growth as they supercharge the existing workforce, delivering unquestionable ROI for coders and healthcare professionals. The world's most important enterprise software platforms like ServiceNow, CrowdStrike, and SAP are integrating NVIDIA's Accelerated Computing and AI stack.

Enterprises broadly are leveraging AI to boost productivity, increase efficiency, and reduce costs. Salesforce's Engineering team has seen at least a 30% productivity increase in new code development after adopting Cursor. This demand spans every market: CSPs, Sovereigns, Model Builders, Enterprises, and Supercomputing Centers, and includes multiple landmark buildouts. Blackwell gained further momentum in Q3 as GB300 crossed over GB200 and contributed roughly 2/3 of the total Blackwell revenue.

What went well
  • Outstanding quarter with revenue of $57 billion, up 62% year-over-year, and record sequential growth of $10 billion (22%).
  • Record Q3 data center revenue of $51 billion, up 66% year-over-year; compute grew 56% while networking more than doubled.
  • Visibility to $500 billion in Blackwell and Rubin revenue from the start of the year through the end of calendar 2026, with room for more orders.
  • GB300 crossed over GB200, contributing roughly two-thirds of total Blackwell revenue, with a seamless transition.
  • Networking revenue of $8.2 billion, up 162% year-over-year, with NVLink, InfiniBand and Spectrum-X Ethernet all contributing; Ethernet GPU attach roughly on par with InfiniBand.
  • Announced AI factory and infrastructure projects aggregating 5 million GPUs across CSPs, sovereigns, model builders, enterprises and supercomputing.
  • Landmark buildouts including xAI's Colossus 2 (first gigawatt-scale data center), Lilly's drug-discovery AI factory, and an expanded AWS/HUMAIN partnership deploying up to 150,000 accelerators.
  • Rubin platform (seven chips) on track to ramp in 2H 2026 with silicon back and bring-up executing well; strategic partnerships with OpenAI (up to 10 GW) and Anthropic announced.
What went wrong
  • Sizable H20 purchase orders never materialized in the quarter due to geopolitical issues and an increasingly competitive China market; H20 sales were only ~$50 million.
  • The current state prevents shipping more competitive data center compute products to China, a market Jensen estimated at ~$50 billion of annual opportunity.
  • Hopper platform, in its 13th quarter, recorded only ~$2 billion in revenue.
  • Persistent supply constraints: the clouds are sold out and the GPU installed base (Blackwell, Hopper, Ampere) is fully utilized, with power, financing, memory and foundry all cited as constraints.

Guidance Changes

MetricPeriodCurrent guidance
Blackwell + Rubin revenue opportunityStart of year through end of CY2026$500B total visibility, on track, with potential upside from new orders (e.g., KSA, Anthropic)
Gross marginQ4 FY2026 and FY2027working to hold gross margins in the mid-70% next year despite known input-price increases
Rubin platform ramp2H FY2027on track to ramp in 2H 2026; silicon received and bring-up executing well

Performance Breakdown

MetricYoYNote
Total revenue +62% $57B on record $10B (22%) sequential growth as customers leaned into accelerated computing, AI models and agentic applications.
Data center revenue +66% Record $51B driven primarily by the GB300 ramp, with compute +56% and networking more than doubling.
Networking revenue +162% $8.2B on NVLink scale-up onset plus double-digit growth in Spectrum-X Ethernet and Quantum-X InfiniBand.
Data center compute +56% Driven primarily by the GB300 ramp crossing over GB200.

Earnings Call Themes & Trends

TopicPrevious mentionCurrent periodTrend
AI bubble concernmarket consternation over buildout magnitude and ROIJensen argued three simultaneous platform shifts (accelerated computing, generative AI, agentic/physical AI) drive real, funded demand; not a bubble
Blackwell GB300 vs GB200GB300 crossing over in Q2GB300 roughly two-thirds of total Blackwell revenue, seamless transition
China / Hopper / H20$2-5B H20 possible in Q2 outlookorders never materialized; H20 only ~$50M, Hopper ~$2B; still advocating to compete in China
CapEx outlook~$600B hyperscaler CapEx (Q2)top-CSP/hyperscaler 2026 CapEx now ~$600B, up more than $200B since the start of the year
Ecosystem investmentsOpenAI relationship since 2016OpenAI (up to 10 GW, potential equity), $10B-scale Anthropic partnership (first time on NVIDIA), xAI/KSA HUMAIN; investing to expand CUDA reach and take equity stakes

Q&A Summary

Is the $500B Blackwell+Rubin (2025-2026) still the parameter, with $150B shipped, and is there upside? (Joe Moore, Morgan Stanley)
Colette confirmed on track toward the $500B forecast, having shipped $50B this quarter, with several quarters remaining through end of CY2026; the number will grow with more orders (e.g., KSA agreement of 400,000-600,000 GPUs over three years, Anthropic), so there is opportunity above $500B.
Is there a realistic path for supply to catch up with demand over 12-18 months? (CJ Muse, Cantor)
Jensen said NVIDIA planned its supply chain well (TSMC, packaging, memory, ODMs) for a big year, driven by three transitions; every layer (accelerated computing, generative AI replacing classical ML, agentic AI) is growing and all run on NVIDIA GPUs, with fastest-growing apps like Cursor, Claude Code and Codex fueling exponential demand.
What NVIDIA content-per-GW assumptions underlie the $500B, and how much of $3-4T needs vendor financing? (Vivek Arya, BofA)
Jensen said NVIDIA's share of the data center rises each generation (Hopper ~$20-25B/GW, Grace Blackwell ~$30B/GW, Rubin higher); perf/watt directly drives customer revenue. Customer financing is up to them; the first two dynamics (accelerated computing, generative AI) are cash-flow funded, with agentic AI as net-new consumption, plus countries and enterprises funding their own.
What are your plans for cash (buyback vs ecosystem investing) and criteria for deals like Anthropic/OpenAI? (Ben Reitzes, Melius)
Jensen said a strong balance sheet underpins supply-chain credibility; NVIDIA will continue buybacks and invests to expand CUDA's ecosystem reach, taking equity in once-in-a-generation companies (OpenAI since 2016; Anthropic's first time on NVIDIA) and expecting extraordinary returns as its platform runs every AI model.
What is the biggest bottleneck that could constrain growth: power, financing, memory or foundry? (Timothy Arcuri, UBS)
Jensen said all are constraints when growing at this scale, but all are tractable and solvable through planning up and down a 33-year supply chain and partnerships in land, power and financing; NVIDIA's best perf/TCO and perf/watt architecture keeps success rate increasing.

More on Nvidia Corp

Reported 2025-11-19 · figures from the Nvidia Corp Q3 2026 earnings call.

See how VectorShift works for your firm

Request Demo