Enterprises are rapidly moving from experimenting with AI in the cloud to deploying full-scale AI factories on-premise. This shift is driven by the need to maintain control over proprietary data, enhance security, and deploy AI agents where the context and action reside.
The discussion frames the next phase of AI as "Agentic AI," where AI systems transition from passive content generators to active agents that can reason, plan, and use digital tools to perform work. This requires a complete reinvention of the computing stack, from GPUs (brains) to new high-performance CPUs (harnesses) and long-term memory systems.
Despite NVIDIA's long-term planning with partners like Micron and SK Hynix, the global supply chain for critical components like HBM memory and advanced node semiconductors cannot keep up with exponential demand. Jensen Huang predicts that even with capacity more than doubling annually, supply will lag demand for at least a decade.
The long-term vision extends beyond the data center. The speakers predict a shift from "personal computers" to "personal AI," with models running locally on devices. They also foresee an even larger market in "physical AI" and robotics, which will bring IT and AI into the $90 trillion of the global economy that is not yet digitized.
The conversation touches on the complex geopolitical landscape, including US export controls on technology to China and the strategic importance of Taiwan as a manufacturing hub. While hopeful for more open markets with China, the speakers acknowledge the reality of restrictions and the ongoing efforts to diversify the supply chain by re-industrializing in the US.
Keep pulling the thread on Michael Dell & Jensen Huang.