Sundar Pichai on Allocating Compute Over Headcount
Sundar Pichai explains how Alphabet budgets granular ML compute and why he reviews team-by-team TPU allocations weekly.
10+ hours of podcasts, in 5 minutes.
Sundar Pichai joins John Collison and Elad Gil to discuss Google's historical and future AI roadmap, including why Google did not ship ChatGPT first despite inventing Transformers. They explore managing Alphabet's $180 billion CapEx budget amid power and memory bottlenecks, the evolution of Search into an agent manager, and long-term moonshots spanning robotics, quantum computing, and space-based data centers.
Sundar Pichai explains how Alphabet budgets granular ML compute and why he reviews team-by-team TPU allocations weekly.
Sundar Pichai and John Collison discuss why AI diffusion stalls inside large enterprises and how startups win the gap before 2027.
Sundar Pichai explains why Google funds space data centers, quantum computing, and robotics with small teams before deploying capital.
Sundar Pichai explains why Google held back LaMDA, how BERT and MUM upgraded Search, and the RLHF bottleneck that delayed a public chatbot launch.
Alphabet projects $185B in CapEx for 2026, but Sundar Pichai says physical limits like memory and electrician shortages will force model compaction.
10+ hours of podcasts, distilled into one 5-minute read. Free, every Sunday.
For now, every subscriber gets both newsletters. No spam. Unsubscribe with one click.