Huang said AI has “moved from testing and evaluating into production” across major enterprises

Eli Lilly (1,000 GPUs on-prem), Samsung, and other large corporates cited as real-world examples

Dell cited 10x–100x productivity improvements versus 10–30% from prior technology cycles

Dell’s role: translating Nvidia technology into enterprise solutions — “Dell will do for the world’s enterprises what the clouds do for the clouds”

Huang introduced the Vera CPU as the highest-performing CPU in the world, designed for agentic AI

Architecture breakdown: GPU brain → Vera CPU harness → NemoClaw agent runtime → OpenShell secure container

Agents need the harness to access memory, use tools, and maintain both working and long-term memory

Dell projects “a lot more CPUs” will be needed as agents use tools “all the time” vs. humans occasionally

Michael Dell estimated hundreds of billions of agents vs. current ~1 billion human digital workers

Biggest constraint: memory and advanced node semiconductors

Huang said demand is “much greater than the overall capacity of the world” but supply chain is “lined up” — HBM, Grace Blackwells, CPUs, silicon photonics all coordinated

Huang told Micron’s Sanjay Mehra and SK Hynix’s Tony the future roadmap “exactly as this is happening right now” three years ago; both are investing

Memory suppliers historically cyclical — Huang argued this is “a complete change in the structure of that market” not a boom-bust cycle

Supply chain is “probably quadrupling every year” but demand growth will keep it tight “for at least a decade”

Huang just returned from China accompanying President Trump

H200 chips are now licensed to sell to China; Huang expressed optimism that market will open further

President Xi stated China will be “an even wider open market”; Premier Li echoed open-market intent

Huang: “I was there to represent the United States and I was honored to do so”

Michael Dell: “I hope there is more economic collaboration between the United States and China”

Neither discussed specifics of H200 sales with Chinese officials directly; that was the president’s purview

Huang: Taiwan “will continue to be one of the epicenters of the world’s technology home”

Nvidia building chip factories, packaging, and AI factories in the United States for supply diversity

Goal is “supply chain diversity and resilience” — not replacing Taiwan but supplementing it

Dell teased “very soon” announcement about collaboration on PCs with powerful Nvidia GPUs

Huang: “Instead of personal computers, we want personal AI”

AI needs to run locally where the context is — laptop for personal use, factory for industrial agents, hospital for medical agents