NVIDIA's Q3 2025 Earnings Call: Contradictions on China Revenue, Supply Chain Shifts, and AI Infrastructure Funding Models

Date of Call: November 19, 2025

Financials Results

• Revenue: $57 billion, up 62% YOY and up 22% sequentially
• Gross Margin: 73.6% (non-GAAP), exceeding outlook, increased sequentially

Guidance:

• Total revenue expected to be $65 billion, plus or minus 2%, implying 14% sequential growth driven by Blackwell momentum.
• GAAP and non-GAAP gross margins expected to be 74.8% and 75%, respectively, plus or minus 50 basis points.
• For fiscal year 2027, working to hold gross margins in the mid-70s.
• GAAP and non-GAAP operating expenses expected to be approximately $6.7 billion and $5 billion, respectively.
• GAAP and non-GAAP tax rate expected to be 17%, plus or minus 1%.

Business Commentary:

Revenue Growth and AI Infrastructure Demand:

• NVIDIA reported record revenue of $57 billion for Q3 2026, up 62% year-over-year and a record sequential growth of $10 billion or 22%.

• The growth was driven by strong demand for AI infrastructure, with significant contributions from the data center segment, particularly from the GB300 ramp and increased compute and networking needs.

Data Center Segment Performance:

• The company's data center revenue reached a record $51 billion, with a 66% year-over-year increase.

• This was attributed to the ongoing transition to accelerated computing, generative AI, and increased utilization of GPU-based workloads across various industries.

Networking Business Growth:

• NVIDIA's networking business, now the largest in the world, generated revenue of $8.2 billion, up 162% year-over-year.

• Growth was driven by the adoption of NVLink, InfiniBand, and Spectrum X Ethernet, which are essential for AI deployments.

Supply Chain and Strategic Partnerships:

• NVIDIA has received silicon back from its supply chain partners and is on track to ramp up the Rubin platform in the second half of 2026.

• Strategic partnerships, such as collaborations with Meta, OpenAI, and Anthropic, are expanding NVIDIA's reach and reinforcing its position in AI infrastructure.

Gaming and Professional Visualization Revenue:

• Gaming revenue was $4.3 billion, up 30% year-on-year, while professional visualization revenue was $760 million, up 56% year-over-year.
• Growth in gaming was driven by strong demand and Blackwell momentum, while professional visualization was driven by the adoption of DGX Spark and AI supercomputers.

Sentiment Analysis:

Overall Tone: Positive

• Management described Q3 as "outstanding" with record revenue growth and expressed confidence in strong demand across multiple transitions. Statements include: "Demand for AI infrastructure continues to exceed our expectations. The clouds are sold out..." and "The world is undergoing 3 massive platform shifts at once... NVIDIA is uniquely addressing each."

Q&A:

• Question from Joseph Moore (Morgan Stanley): Update on the $500 billion Blackwell plus Rubin revenue forecast for 2025/2026 and potential upside.
  Response: Confirmed on track for $500 billion through end of calendar 2026, with additional orders expected from new deals like KSA and Anthropic.

• Question from Christopher Muse (Cantor Fitzgerald): Realistic path for supply to catch up with demand over next 12-18 months given sold-out conditions and multiple technology transitions.
  Response: Confident supply chain planning with global partners is strong, citing the convergence of three exponential platform shifts driving demand.

• Question from Vivek Arya (Bank of America Securities): Assumptions on NVIDIA content per gigawatt in the $500 billion number and the portion of the $3-4T AI build by 2030 requiring vendor financing.
  Response: Content per gigawatt increases each generation (~$30B for Blackwell). Hyperscaler investments in accelerated computing and generative AI are cash flow funded; new agentic AI and other industries will fund their own builds.

• Question from Benjamin Reitzes (Melius): Plans for significant future cash generation, including buybacks vs. ecosystem investments like those in OpenAI and Anthropic.
  Response: Cash used to secure resilient supply chain and support growth; will continue stock buybacks. Ecosystem investments expand CUDA reach and offer stakes in high-growth companies.

• Question from James Schneider (Goldman Sachs): Expectation for AI inference shipment percentage next year and context on Rubin CPX product and target applications.
  Response: Hopes inference becomes a large market, indicating heavy usage. Rubin CPX is designed for long-context workloads with excellent performance-per-dollar.

• Question from Timothy Arcuri (UBS): Single biggest bottleneck (power, financing, memory, foundry) that could constrain growth.
  Response: All are constraints but manageable with planning and partnerships; confident in NVIDIA's architectural value and growing customer success.

• Question from Stacy Rasgon (Bernstein Research): Biggest cost increases impacting margins and outlook for OpEx growth next year.
  Response: Input costs are rising but working through cost improvements, cycle time, and mix to hold gross margins in the mid-70s. OpEx will grow to support innovation for new architectures.

• Question from Aaron Rakers (Wells Fargo): Role of AI ASICs/XPUs in architecture build-outs given recent deals and if stance has changed.
  Response: Argues NVIDIA's versatility across every phase of AI transition, every AI model, every cloud/on-prem deployment, and strong offtake makes it superior to narrow ASICs.

Contradiction Point 1

Revenue Forecast Certainty and Underlying Drivers

The $500 billion forecast shifts from a broad multi-year opportunity to a specific, trackable annual target driven by new orders, altering its perceived immediacy and certainty.

Could you outline your growth outlook through 2026 and provide the Network vs. Data Center revenue split? - Christopher Muse (Cantor Fitzgerald & Co.)

2026Q2: The company is on track to scale into a $3 trillion to $4 trillion AI infrastructure opportunity over the next five years... Visibility is strong based on forecasts from large customers and the explosive growth of AI-native start-ups... - Jen-Hsun Huang(CEO)

Can you update on the $500 billion Blackwell and Rubin revenue visibility for 2025/2026, specifically the $350 billion remaining after the initial $150 billion shipment? Is there potential for upside demand? - Joseph Moore (Morgan Stanley)

20251120-2026 Q3: Yes, the company is on track for the $500 billion forecast through the end of calendar 2026. New orders are expected, as evidenced by recent announcements... - Colette Kress(CFO)

Contradiction Point 2

Characterization of China Market Opportunity and Constraints

The narrative shifts from a clear near-term ~$50B opportunity limited by export controls to one dependent on unresolved geopolitical issues, adding uncertainty and risk to a key growth market.

What are the long-term prospects for China? How important is Blackwell architecture licensing in China? - Joseph Lawrence Moore (Morgan Stanley)

2026Q2: China is the second-largest computing market... NVIDIA sees a ~$50B opportunity in China this year and is advocating for American companies to be able to address this market. - Jen-Hsun Huang(CEO)

What needs to be done to achieve the potential $2B-$5B in Q3 shipments to China, and what is the sustainable pace for the China business into Q4? - Vivek Arya (Bank of America Securities) - Part 1 of his question in 2026Q2

20251120-2026 Q3: The shipment amount depends on the resolution of ongoing geopolitical issues between governments and companies. NVIDIA has ready supply and could ship up to $5B if more licenses are issued. - Colette Kress(CFO)

Contradiction Point 3

Primary Growth Driver for Data Center Revenue

The narrative shifts from a singular AI driver needing more compute to three concurrent platform shifts, suggesting a change in management's growth strategy due to market dynamics or competition.

Given 12-month wafer-to-rack lead times and Rubin's second-half ramp confirmation, what is your growth outlook for 2026 and how will Network and Data Center segments be split? - Christopher Muse (Cantor Fitzgerald & Co.)

2026Q2: The primary growth driver is the evolution to reasoning agentic AI, which requires orders of magnitude more compute than simple chatbots. - Jen-Hsun Huang(CEO)

Given current GPU shortages and the introduction of new AI models like Gemini 3 and Rock 5, what is the realistic path for supply to catch up with demand over the next 12-18 months, or could it take longer? - Christopher Muse (Cantor Fitzgerald)

20251120-2026 Q3: Demand is driven by three concurrent platform shifts: (1) general-purpose to accelerated computing, (2) classical to generative AI, and (3) the rise of agentic AI. All these applications run on NVIDIA GPUs... - Jen-Hsun Huang(CEO)

Contradiction Point 4

Supply Chain Constraint Impact

The shift from acknowledging supply constraints limiting growth to stating they are under control marks a critical change in risk assessment and guidance.

Would Q4 guidance be higher without China restrictions? Are you supply constrained, and are lead times decreasing from redirected China demand? - Stacy Rasgon (Bernstein Research)

2024Q3: Yes, Q4 guidance could have been higher without China restrictions, as some supply not constrained could have been sold to China. - Colette Kress(CFO)

With current GPU shortages and high utilization from new AI models like Gemini 3 and Rock 5, how long will it take for supply to meet demand—12-18 months or longer? - Christopher Muse (Cantor Fitzgerald)

20251120-2026 Q3: NVIDIA's supply chain planning with partners (TSMC, memory vendors, ODMs) is strong... Demand is growing across three concurrent platform shifts, and the company has established routes to market... - Jen-Hsun Huang(CEO)

Contradiction Point 5

Nature of Hyperscaler Funding for AI Infrastructure

The shift from hyperscaler investments being "cash-flow funded" to introducing "net consumption" for agentic AI suggests a change in demand drivers and financial risk.

What assumptions underpin the $500 billion figure for NVIDIA content per gigawatt? How much of the $3-4 trillion data center TAM by 2030 will be financed by vendors versus self-funded by customers/governments? - Vivek Arya (Bank of America Securities)

20251120-2026 Q3: Hyperscaler investments in accelerated computing and generative AI are cash-flow funded as they improve existing business models (e.g., recommender systems). - Jen-Hsun Huang(CEO)

What assumptions underpin the $500B figure for NVIDIA's content per gigawatt? How much of the $3-4T data center TAM by 2030 will rely on vendor financing versus customer cash flows? - Vivek Arya (Bank of America Securities)

2026Q3: Agentic AI represents new, net consumption and new applications, also growing rapidly. - Jen-Hsun Huang(CEO)