What does Nvidia do and what are its main business segments?

What factors drove the changes in Nvidia's revenue and profit?

What is the revenue and EPS growth rate for Nvidia year over year?

What is Nvidia's latest dividend and current dividend yield?

What guidance did Nvidia's management provide for the next earnings period?

Did Nvidia beat or miss consensus estimates last quarter?

What is Nvidia's gross profit margin?

What is the market's earnings forecast for Nvidia next quarter?

- **Record Growth**: Q3 revenue $57B (+62% YoY), data center $51.2B (+66%), networking $8.2B (+162%).
- **Pipeline Expansion**: $500B pipeline (FY25-CY26), $50B shipped Q3, with new deals like KSA's 400k-600k GPUs.
- **AI Leadership**: CUDA ecosystem drives 10x performance/watt, A100 GPUs still running 6 years later. Partnerships with Anthropic (1 GW compute) and OpenAI (AI factory).
- **Supply Chain Resilience**: U.S. manufacturing, 33-year partner network, and 10x Rubin performance over Blackwell.
- **Cash Strategy**: $0.5T free cash for buybacks, ecosystem investments (e.g., Anthropic, OpenAI), targeting mid-70s gross margins.

**【CFO Financial Commentary】**
- Q3 revenue: $57 billion (62% YoY growth), with data center revenue at $51.2 billion (66% YoY). Networking revenue surged 162% YoY to $8.2 billion.
- $500 billion pipeline from FY25 to CY26, with $50 billion shipped in Q3. Gross margins: 73.6% (non-GAAP), targeting mid-70s for FY27.
- Rubin platform on track for H2 FY26, offering 10x performance improvement over Blackwell.

**【CEO Remarks】**
- Three platform shifts: accelerated computing, generative AI, and agentic AI. NVIDIA's CUDA ecosystem enables all phases of AI, from pre-training to inference.
- Strategic partnerships with Anthropic (1 GW compute commitment) and OpenAI (AI factory collaboration). CUDA compatibility extends GPU lifecycle, e.g., A100 GPUs still running at full utilization.
- Supply chain resilience: U.S. manufacturing of Blackwell wafers with TSMC, expanding U.S. presence with Foxconn and others.

**【Q&A Highlights】**
- **Pipeline Expansion**: Colette confirms $500 billion pipeline, with new deals like KSA's 400k-600k GPUs over 3 years. Anthropic's 1 GW compute adds to growth.
- **Supply Chain Confidence**: Jensen highlights NVIDIA's architecture leadership (10x performance/watt over H200) and supply chain resilience, securing 33-year partners.
- **Cash Utilization**: $0.5T free cash flow to be used for buybacks, ecosystem investments (e.g., Anthropic, OpenAI), and growth. Investments expand CUDA reach, e.g., OpenAI's 1M enterprise customers.
- **Margin Management**: Colette targets mid-70s gross margins via cost improvements and mix optimization, despite rising input costs.

NVIDIA Corporation (NVDA) Q3 2026 Earnings Call Transcript

- **Record Q2 Growth**: $46.7B revenue (+56% data center YoY), driven by Blackwell (GB300 at 1,000 racks/week) and sovereign AI ($>2B Q2).
- **AI Infrastructure Leadership**: Targeting $3-4T global spend by 2030, with NVIDIA capturing $35B+/gigawatt. Rubin platform in fab for 2026 launch.
- **China Opportunity**: $50B+ annual potential if H20 approved; 50%+ CAGR driven by open source models and enterprise AI.
- **Networking Dominance**: $7.3B networking revenue (+98% YoY), Spectrum-XGS doubles GPU-GPU speed, NVLink 144 offers 14x PCIe bandwidth.
- **Sovereign AI Expansion**: $20B+ annualized revenue target, with UK/EU AI factories and 21 exaflops UK supercomputer.

**【CFO Financial Commentary】**
- Q2 revenue: $46.7 billion (YoY +56% in data center). Blackwell platform revenue up 17% sequentially, GB300 production at 1,000 racks/week.
- GAAP/non-GAAP gross margin: 72.4%/72.7%, driven by $180M inventory release. Sequential OpEx growth due to compute infrastructure costs.
- Q3 outlook: $54B ±2% revenue, 73.3-73.5% gross margin. H20 China shipments excluded; potential $2-5B if licenses approved.
- Share buybacks: $10B returned in Q2; $60B new authorization approved.

**【CEO Strategic Remarks】**
- AI infrastructure spend to reach $3-4T by 2030, with NVIDIA targeting $35B+ per gigawatt factory. Rubin platform in fab, expected 2026 launch.
- Blackwell Ultra (GB300) delivers 10x token efficiency vs Hopper, 50x energy efficiency. NVFP4 4-bit precision enables 7x faster training.
- Networking revenue: $7.3B (YoY +98%), Spectrum-XGS Ethernet doubles GPU-GPU speed. NVLink 144 switch bandwidth 14x PCIe Gen5.
- Sovereign AI: $>2B revenue Q2, targeting >$20B annualized. UK, EU AI factories planned with 20 exaflops+ systems.

**【Q&A Highlights】**
- **China Market**: $50B+ annual opportunity if H20 approved; 50%+ CAGR potential. Open source models (e.g., DeepSeek) drive enterprise adoption.
- **ASIC Competition**: NVIDIA's full-stack solution (GPU+CPU+NIC+switch) offers 100x better performance/watt. 90+ companies adopt RTX PRO servers.
- **Long-term Growth**: Rubin to enable multi-gigawatt factories. AI-native startups funded at $180B YTD, 10x revenue growth vs 2024.
- **Power Constraints**: Per Jensen, 60% of global compute in US; perf/watt directly impacts factory revenue. Rubin to improve efficiency further.

NVIDIA Corporation (NVDA) Q2 2026 Earnings Call Transcript

- **Record Growth**: Q1 revenue $44B (+69% YoY), Data Center leads at $39B (+73%).
- **AI Inference Explosion**: 100 trillion tokens processed, 72,000 GPUs/week shipped to hyperscalers.
- **China Risk**: $50B AI market inaccessible, $4.5B inventory/write-down from H20 export ban.
- **Sovereign AI Surge**: 100+ AI factories in development, UAE/Saudi Arabia projects announced.
- **Enterprise AI Breakthrough**: RTX Pro Enterprise, DGX Spark/Station enable on-prem AI at scale.
- **Networking Dominance**: Spectrum-X annualizing $8B+, NVLink72 shipments >$1B/Q1.

**【CFO Financial Highlights】**
- Q1 2026 revenue: $44 billion (69% YoY growth), driven by Data Center ($39B, +73%) and Gaming ($3.8B, +42%).
- H20 export controls impact: $4.5B inventory/write-down charge, $2.5B unshipped revenue, and $8B expected Q2 China Data Center loss.
- Blackwell ramp success: 70% of Data Center compute revenue, 72,000 GPUs/week shipped to hyperscalers.
- Networking growth: $5B Q1 (64% QoQ), Spectrum-X annualizing $8B+ with Google Cloud/Meta additions.

**【CEO Strategic Priorities】**
- **AI Inference Surge**: 100 trillion tokens processed in Q1 (+5x YoY), Microsoft/AI factories deploying 1,000+ NVL72 racks/week.
- **Sovereign AI Expansion**: 100+ AI factories in development, including Saudi Arabia (500MW), UAE (5GW), Taiwan, Sweden.
- **Enterprise AI Products**: RTX Pro Enterprise, DGX Spark (1 petaflop), DGX Station (20 petaflops) for on-prem AI.
- **China Market Risk**: $50B TAM inaccessible due to H20 export ban, $4.5B inventory/write-down.

**【Q&A Key Takeaways】**
- **Inference Growth**: Reasoning models require 1,000x more tokens, Grace Blackwell 40x faster than Hopper.
- **Supply Chain**: Onshore manufacturing (Arizona fabs, Texas supercomputer plants) ramping to meet demand.
- **Sovereign Deals**: UAE/Saudi Arabia projects announced, more sovereign AI factories in pipeline.
- **China Recovery**: No current workaround for H20, exploring limited options for compliant products.
- **Networking Momentum**: Spectrum-X adoption rate 3.5x faster than prior gen, NVLink72 shipments >$1B/Q1.

NVIDIA Corporation (NVDA) Q1 2026 Earnings Call Transcript

- **Record Q4 Growth**: $39.3B revenue (+78% YoY), driven by $11B Blackwell sales and 93% YoY Data Center growth.
- **Blackwell Dominance**: Fastest product ramp in company history, with 100,000+ GPU clusters shipped and 25x faster reasoning AI performance vs Hopper.
- **Gross Margin Outlook**: Mid-70s by FY2026; Q1 2026 margin at 70.6% (non-GAAP).
- **Future Products**: Blackwell Ultra (Q2 2025), Vera Rubin, and COSMO World physical AI platform.
- **Enterprise AI Expansion**: 5.9M developers, 90,000+ Llama derivatives, and 103% YoY Automotive revenue growth.
- **Cost Efficiency**: NVIDIA tools reduce inference costs by up to 3x (e.g., Perplexity, Microsoft Bing).
- **Risks**: China export controls limit Data Center growth, but U.S. demand offsets with 16% Q4 sequential growth.

**【CFO Financial Commentary】**
- Q4 2025 revenue: $39.3 billion (12% sequential growth, 78% YoY), exceeding guidance of $37.5 billion.
- FY2025 total revenue: $130.5 billion (114% YoY growth).
- Data Center revenue: $115.2 billion (100% YoY growth), driven by Blackwell sales ($11B in Q4) and Hopper 200 growth.
- Blackwell ramp: Fastest product launch in NVIDIA history, with 100,000+ GPU clusters shipped.
- Gross margin: 73.5% (non-GAAP), expected to improve to mid-70s by FY2026 as Blackwell ramps.
- Q1 2026 outlook: $43B revenue (+/-2%), sequential growth in Data Center and Gaming.

**【CEO Jensen Huang Commentary】**
- AI scaling laws: Pre-training, post-training (reinforcement learning), and inference (reasoning AI) driving compute demand.
- Blackwell performance: 25x faster token throughput vs Hopper 100, 20x lower cost for reasoning AI.
- Future products: Blackwell Ultra (Q2 2025), Vera Rubin, and physical AI platforms like COSMO World.
- Enterprise AI adoption: 5.9M developers using NVIDIA tools, 90,000+ Llama derivatives on Hugging Face.
- Automotive growth: FY2025 revenue $1.7B (103% YoY), Toyota, Aurora, and Continental partnerships.

**【Q&A Highlights】**
- **Gross Margin Trajectory**: Colette expects mid-70s gross margin by FY2026 as Blackwell ramps; Q1 2026 margin at 70.6% (non-GAAP).
- **Blackwell Ultra Readiness**: Jensen confirms on-track for H2 2025 launch, with seamless transition from current Blackwell systems.
- **ASIC vs GPU Debate**: NVIDIA's general-purpose architecture, ecosystem, and performance per watt优势使GPU成为AI基础设施首选。
- **China Market Dynamics**: China data center revenue remains below pre-export control levels; NVIDIA compliant with regulations but cautious on near-term growth.
- **AI Factory Economics**:推理计算成本两年下降200倍，NVIDIA通过TensorRT、Triton等工具帮助客户实现TCO优化（如Perplexity推理成本降低3x）。


NVIDIA Corporation (NVDA) Q4 2025 Earnings Call Transcript

NVIDIA (NVDA)-20 Nov 24-2025 Q3 Earnings call transcript

NVIDIA (NVDA)-28 Aug 24-2025 Q2 Earnings call transcript

NVIDIA (NVDA)-22 May 24-2025 Q1 Earnings call transcript

NVIDIA (NVDA)-21 Feb 24-2024 Q4 Earnings call transcript

- **Record Q3 Revenue**: $18.1 billion (34% sequential, 200%+ YoY), driven by data center ($14.5B, 279% YoY) and networking ($>10B annually). 
- **AI Leadership**: H100 GPUs dominate AI training; TensorRT LLM + H200 reduce inference costs by 4x. Grace Hopper systems deployed in U.K. and German supercomputers.
- **China Impact**: Q4 data center revenue from China expected to decline 20-25% due to export controls, but growth elsewhere will offset.
- **New Products**: 
  - **H200 GPU**: 2x inference performance over H100, supported by major clouds.
  - **Spectrum-X Ethernet**: 1.6x AI performance boost for enterprise AI.
  - **Grace Hopper**: Targeting multibillion-dollar HPC/AI market.
- **Growth Drivers**: 
  - **CPU (Grace)**: Expanding into data center compute.
  - **Networking**: InfiniBand for AI factories, Spectrum-X for enterprise.
  - **Software**: NVIDIA AI Enterprise ($1B+ annualized run rate).

**【Management Opening Remarks】**
- Simona Jankowski welcomed attendees and provided forward-looking statements, emphasizing the risks of forward-looking statements and non-GAAP financial measures. She introduced Jensen Huang and Colette Kress, then handed over to the CFO.

**【CFO Financial Commentary】**
- **Q3 Financial Highlights**: Revenue of $18.1 billion (34% sequential growth, 200%+ YoY), with data center revenue at $14.5 billion (41% sequential, 279% YoY). Networking revenue exceeded $10 billion annually, up fivefold YoY.
- **Product Performance**: H100 GPUs dominate AI training; TensorRT LLM and H200 reduce inference costs by 4x. Grace Hopper systems are being deployed in supercomputers globally.
- **Export Controls Impact**: China and other destinations contributed 20-25% of data center revenue but will decline significantly in Q4 due to U.S. export restrictions. New compliant products are in development but will not materially impact Q4.
- **Q4 Guidance**: Revenue expected at $20 billion (±2%), with data center growth driven by compute and networking. Gross margins projected at 74.5% (GAAP) and 75.5% (non-GAAP).

**【CEO Commentary】**
- **Generative AI Adoption**: Generative AI is transforming industries, with NVIDIA's HGX platform (Hopper + InfiniBand) as the reference architecture for AI supercomputers. Enterprise AI adoption is accelerating, with companies like Adobe, SAP, and ServiceNow integrating AI copilots.
- **Product Innovations**: 
  - **Grace Hopper**: In production, targeting HPC and AI factories. U.K. and Germany are building supercomputers with Grace Hopper systems.
  - **H200 GPU**: Delivers 2x inference performance over H100, with support from major cloud providers.
  - **Spectrum-X Ethernet**: Optimized for AI workloads, launched with Dell, HPE, and Lenovo.
- **New Growth Drivers**: 
  - **CPU (Grace)**: Expanding into data center compute with Grace Hopper systems.
  - **Networking**: InfiniBand for AI factories, Spectrum-X for enterprise AI.
  - **Software & Services**: NVIDIA AI Enterprise (annualized $1B+ run rate) and DGX Cloud for custom AI model development.
- **Sovereign AI Infrastructure**: Partnerships with India, France, and others to build national AI clouds, representing a multibillion-dollar opportunity.

**【Q&A Highlights】**
- **China Export Controls**: Q4 data center revenue from China expected to decline 20-25%, offset by growth elsewhere. New compliant products will be introduced but not material in Q4.
- **Grace Hopper Production**: In high-volume production, targeting multibillion-dollar product line. European supercomputers are early adopters.
- **Networking Expansion**: Spectrum-X Ethernet solutions are being adopted by enterprise customers through partnerships with Dell, HPE, and Lenovo.
- **AI Foundry Service**: Monetized through NVIDIA AI Enterprise licenses ($4,500/GPU/year). Partners like Microsoft, SAP, and ServiceNow are using the service to build custom AI models.
- **Data Center Growth Outlook**: Expect continued strong growth through 2025, driven by supply expansion, new customers (sovereign AI clouds, enterprise AI), and product innovation (Grace, H200, Spectrum-X).

NVDA

NVDA fundamentals

Earnings Call

You can ask Aime