Etched says inference means owning the full stack, not just writing kernels
AccBalanced · x · 2026-07-25
Etched says its inference stack goes far beyond kernel writing: the company claims ownership of the full stack, from KV cache management and sharding algorithms to workload-aware SerDes tuning.
It says it pushes the Pareto frontier daily, co-designs models with customers, works with chip designers, and continuously improves its model harness for recursive kernel development. The pitch also highlights topics like KV tree masking, diffusion speculators, latent compression, and large-scale scale-up domains.
More from coding & agent
- Statistician Shares AI Workflow: Let AI Handle Narrative, Humans Keep the Math — Excellent_Copy4646 · 2026-07-25
- GitHub screenshot jokes about 41 subagents and 157,288 new lines of code — zeeg · 2026-07-25
- OpenAI adds voice to desktop ChatGPT so users can steer Work and Codex by speaking — 新智元 · 2026-07-25
- Anthropic makes agent memory file-based, exportable, auditable, and reversible — VraserX · 2026-07-25
- “Typing code was never the job,” argues a post on Claude and $500K engineers — gdechichi · 2026-07-25
- Open-source TokenShield adds turn-scoped hashes to stop CrewAI retry loops — bulleykebaal · 2026-07-25