Nvidia CEO Jensen Huang on Blackwell Sales, Vera Rubin, and China Market

Blackwell and Vera Rubin Platforms

• 🚀 Blackwell GPUs are experiencing sales that are "off the charts," with Nvidia's cloud offerings currently sold out.
• 💡 Nvidia has planned its supply chain effectively, with partners like TSMC and Micron ensuring ample Blackwell and Vera Rubin production.
• 🎯 The Vera Rubin platform is on track for delivery around Q3 of next year, continuing Nvidia's annual product cycle, and is expected to be a significant success.
• 🧩 The rack scale architecture, including the MVLink 72 switch, is a revolutionary system that seamlessly transitions from Grace Blackwell to Vera Rubin, ensuring a smooth ramp-up.

China Market Strategy

• ⚠️ Nvidia's forecast for the China market is currently zero, despite its importance.
• 💬 The company aims to re-engage the Chinese market with excellent products and compete globally, emphasizing the benefits for both the US and China.
• 🤝 Nvidia is committed to engaging with both the US and Chinese governments to encourage participation in the open market, while adhering to current restrictions.

Energy and Infrastructure Constraints

• ⚡ The rapid growth of Nvidia, with quarterly growth of $10 billion, presents challenges across the entire supply chain, including energy providers.
• 🔌 Nvidia leverages its extensive go-to-market network, including every cloud service provider and OEM, to find available power sources globally.
• 🧠 The company's architecture runs all major AI models, ensuring high demand and enabling deployment across various cloud environments.

AI Compute Demand and Investment

• 📈 Nvidia is disciplined in its investments and build-out, aligning with the visibility of demand and financing capabilities of partners like OpenAI and Anthropic.
• 🚀 OpenAI and Anthropic are experiencing exponential growth in compute demand and adoption, with their engineering teams working to scale capacity and optimize usage.
• 🌐 The increasing number of AI applications necessitates a corresponding exponential growth in compute demand.

Chip Longevity and Value

• 💡 Nvidia's CUDA architecture provides versatility, allowing GPUs to accelerate all phases of AI (pre-training, post-training, inference) and run diverse models.
• ⏳ Older chips like the Ampear A100, shipped six years ago, remain highly valuable due to continuous software updates and their superior performance compared to CPUs for many workloads.
• 📉 Unlike singular-use accelerators, Nvidia GPUs maintain their value over time due to their broad applicability and versatility.

Financial Contributions

• 💰 Nvidia's contribution to a 1-gigawatt data center is approximately $35 billion for a Vera Rubin system, representing about 35% of the total cost.