【Earnings Highlights and Outlook】
- Q4 Revenue: $39.3 billion, up 12% sequentially and 78% year-on-year, exceeding the outlook of $37.5 billion.
- FY2025 Revenue: $130.5 billion, up 114% from the prior year.
- Q1 FY2026 Revenue Outlook: $43 billion (±2%), with expected sequential growth in both data center and gaming segments.

【Q&A Highlights】

Q1: How is the demand for large language models (LLMs) evolving, and what does this mean for inference-dedicated clusters?
A: Jensen Huang highlighted three scaling laws driving AI computing demand:
1. Traditional scaling for foundation models and multi-modality.
2. Post-training scaling (reinforcement learning, fine-tuning) requiring more compute than pre-training.
3. Test-time scaling or "reasoning AI," where inference compute needs are 100x more than one-shot examples.
- Blackwell architecture is designed to handle all these scaling dimensions efficiently.
- The future may see hundreds of thousands to millions of times more compute for advanced reasoning models.
- Nvidia's architecture remains popular due to its flexibility across various AI workloads and configurations.

Q2: What's the status of GB200 (Blackwell) ramp-up and the transition to future generations?
A:
- Blackwell ramp-up is successful, with $11 billion in revenue last quarter.
- The transition from Blackwell to Blackwell Ultra will be smoother than Hopper to Blackwell, as the system architecture remains the same.
- Blackwell Ultra is on track for the second half of the year.
- Future generations (e.g., Vera Rubin) are already being prepared with partners.

Q3: How does Nvidia view the balance between custom ASICs and merchant GPUs?
A: Jensen Huang emphasized Nvidia's advantages:
1. General-purpose architecture optimized for various AI models and workloads.
2. End-to-end capabilities from data processing to inference.
3. Widespread availability across clouds and on-premises.
4. Rapid performance improvements, directly translating to revenue for data centers.
5. Robust software stack and ecosystem, which is challenging to replicate.

Q4: How is demand distributed geographically, especially with potential regulations affecting certain regions?
A:
- China's revenue remains at approximately half of what it was before export controls.
- AI has become mainstream across various industries and geographies.
- Nvidia believes the AI transition is still in its early stages, with significant growth potential across different regions and sectors.

Q5: How is enterprise demand evolving within the data center segment?
A:
- Enterprise business grew 2x year-on-year, similar to large CSPs.
- Long-term, enterprise and industrial AI applications are expected to be larger than CSP-driven demand.
- Nvidia sees three key areas of AI compute in enterprises:
1. Agentic AI for employee productivity and company operations.
2. AI factories for training and improving products (e.g., autonomous vehicles).
3. Physical AI for robotics and industrial applications.

Overall, Nvidia remains highly optimistic about sustained demand for AI computing, driven by advancements in reasoning AI, physical AI, and the broader adoption of AI across industries.
Disclaimer: This earnings call summary is generated by AI and is for informational purposes only. Due to technical limitations, inaccuracies may exist. It does not constitute investment advice or commitments.