Silicon Data pushes back: falling token index means cheaper mix, not stalled AI demand
AccBalanced · x · 2026-09-24
Responding to Fidelity's Timmer claim that token expenditures and GPU lease rates are flat-to-down, Silicon Data corrects the record: its Token Index is an expenditure-weighted price per million tokens, not spending — expenditures and volumes are growing exponentially, and the falling index just reflects a cheaper mix (routing, cheaper tasks, flash models). H100/A100 rents did flatten, but they're no longer the binding market: since June 21, B300 rents are up 37% and B200 up 8%. Silicon Data reads this as mix-shift and a Jevons effect — quantity growing far faster than prices fall — not a demand stall.
More from Infra
- Anthropic paid $5B for short-term compute in Q3 as spot prices run ~2x long-term deals — tengyanAI · 2026-11-11
- UK Sovereign AI invests in Fractile to build next-gen frontier AI chips — charlieharris01 · 2026-11-11
- What happened when I moved vector retrieval into Snowflake SQL instead of orchestrating RAG outside — dima806_dima · 2026-11-11
- AI chip startup Fractile raises $720M Series C for DRAM-on-processor design — Tom_Westgarth15 · 2026-11-11
- Mobile AI Factory: agentic pipeline builds on-device SLMs that beat 5-shot frontier models — kalyan_kpl · 2026-11-11
- Vivo's Starlink direct-to-cell service hits 10M users in two weeks across Brazil — XFreeze · 2026-11-11