AFFN disaggregation is the holy grail of inference networking, says Gavin Baker
GavinSBaker · x · 2026-08-24
Investor Gavin Baker argues that AFFN disaggregation — splitting attention/FFN inference stages across different hardware — is the holy grail, but an extremely hard networking problem.
The post he replies to describes the new "Pareto optimal" scheduling: agent CPU control & tool-call scheduling, GPU prefill (KV cache), and ASIC FFN decode (e.g. Cerebras). It further claims AWS is technically prepared for this era, other hyperscalers are lagging, and no NeoCloud or inference leader today comes close.
Related event: Gavin Baker Predicts 100x AI Spend Growth and a Minimum Token-Spend Barrier(5 posts)→
More from Infra
- Analyst: 80% of neoclouds may not survive, but the demand for their GPU capacity is real — BenBajarin · 2026-10-16
- Apple and Meta partner to run Muse Spark model on linked Mac Studios — pdamodaran · 2026-10-16
- Nvidia's developer base hits 11 million, up ~3.7x from 3 million in 2021 — Beth_Kindig · 2026-10-16
- Muse Spark Open runs locally on a Mac M3 Ultra via llama.cpp: frontier model on desktop — pcuenq · 2026-10-16
- Muse-Spark open-weights drop: 816B-A42B MoE model gets 359GB MLX 3-bit build — ns123abc · 2026-10-16
- Meta launches Muse Spark Open, says it runs on two M5 Ultra Mac Studios with Apple's help — The Verge AI · 2026-10-16