Join us as we unpack the NVIDIA Rubin platform—the next-gen AI supercomputer built around extreme co-design. We map the six-chip system (Vera CPU, Rubin GPU, NVLink 6, Connect X9, BlueField 4 DPU) and its groundbreaking bandwidth and efficiency gains, from 3.6 TB/s per GPU to 10x lower inference costs and MOE training with 4x fewer GPUs. We explore the Inference Context Memory Storage Platform, the rise of agentic AI, and what Rubin means for scaling confidential AI at the forefront of industry adoption by AWS, Google, Microsoft, and major AI labs. A Vera Rubin-inspired look at the invisible infrastructure powering the future of AI.

Note: This podcast was AI-generated, and sometimes AI can make mistakes. Please double-check any critical information.

Sponsored by Embersilk LLC