Google Exec: 7-Year-Old TPUs Still Running at 100% Utilization
rohanpaul_ai · x · 2026-08-13
Amin Vahdat, GM of AI and Infrastructure at Google Cloud, revealed that Google's TPUs from 7 or 8 years ago are still seeing 100% utilization today.\n\nThis illustrates the Jevons Paradox in full swing across the AI compute landscape: as hardware becomes more efficient and daily breakthroughs occur, overall demand for compute explodes rather than shrinks. Every available chip is currently being put to work to meet the massive scaling needs of AI.
More from Infra
- New hardware day: local LLM deployment the moment it arrives — lxfater · 2026-10-16
- Together AI's score centering fixes RL training collapse under quantized inference — nrehiew_ · 2026-10-16
- Inference is splitting prefill and decode across specialized chips, not just H100s — AccBalanced · 2026-10-16
- Rebasely launches in beta: Git hosting built for humans and coding agents — handotdev · 2026-10-16
- Drama3 TTS lands in audio.cpp: up to 5x real-time on RTX 5090, <8GB VRAM mode too — Acceptable-Cycle4645 · 2026-10-16
- Three CoreWeave co-founders have sold $3.3B in shares since IPO amid credit worries — SumitGup · 2026-10-16