The Great "Rack Spread": B300 GPU rents up 106% while token prices fall 58%
FinanceYF5 · x · 2026-11-01
Per Ornn data cited by a16z (as of Oct 26, 2026), an inflection called the "Rack Spread" is underway: token prices keep falling while GPU rental rates climb.
Over the 12 weeks since mid-July: B300 rental rates rose 106%, B200 +65.7%, H100 +16.9%, while blended token cost fell 58.4%. The driver: rapidly growing inference demand against still-tight supply of high-end compute, splitting GPU rents and model API pricing in opposite directions.
More from Infra
- Leaked: Meta has ~1.3M GPUs with 40% sitting idle, offloading capacity that threatens neoclouds — SumitGup · 2026-11-01
- Picked up a $1,100 RTX 3090 FTW3 for a headless local LLM server — seeking configs — NighthawkXL · 2026-11-01
- Sequoia's Shaun Maguire builds SpaceX model projecting $731B revenue by 2028 — FinanceYF5 · 2026-11-01
- SK hynix sold out through 2029: AI selloff is about cash flow, not demand, argues Puru Saxena — danielrock · 2026-11-01
- Gemma 4 reportedly running in a space data center under Google's Project Suncatcher — Saboo_Shubham_ · 2026-11-01
- Graphics compilers trail nvcc by orders of magnitude on compute code quality — mike64_t · 2026-11-01