This episode is available to stream on-demand.
Scaling AI workloads requires more than just adding more GPUs into racks. The entire facility, from chip design to cooling architecture, must evolve to accommodate densities that challenge traditional infrastructure. Collaboration with leading technology partners like NVIDIA is essential, as their innovation at the silicon and system level drives new requirements for power, thermal design, and real-time performance optimization. This session dives into:
- Engineering for extreme power densities, beyond 100kW per rack
- The impact of AI model scaling on facility power distribution
- How modularity and prefabrication accelerate AI deployment
- Technical deep dive into AI workloads
- Chip innovation, and performance optimization
