Jensen Huang: Hyperscalers Plan Once a Year and Are 'Always Almost Wrong' on AI
rohanpaul_ai · x · 2026-09-16
Speaking on the All-In Podcast, Jensen Huang called out a major weakness of hyperscalers: their infrastructure planning cadence is too slow for AI.
- His point: "The hyperscalers plan once a year, but the market dynamics is so volatile right now that they're always almost wrong."
- He argues this is exactly where agile neo-clouds have an advantage.
The remark frames AI compute supply-demand mismatch as a planning-cycle problem at big clouds rather than a chip-capacity problem.
More from Companies & People
- Rob Mulla marks one year at Google scaling TPU inference with vLLM — Rob_Mulla · 2026-09-16
- Meta One Advanced starts at $49.99/month, selling post and Reels hyperlinks as a paid feature — HaktanSuren · 2026-09-16
- Jensen Huang says 'Just follow me,' Salesforce's Benioff replies 'We're all following you, bro' — kimmonismus · 2026-09-16
- Salesforce launches Koa, its first CRM reasoning model, claiming 3x fewer errors — LiJunnan0409 · 2026-09-16
- PyMC Labs opens free contributor seats for its final 2026 Agentic Data Science cohort — twiecki · 2026-09-16
- Recruiter: 30/60/90-Day Plans Are Dead, Everything Has Accelerated — morganb · 2026-09-16