This page may contain stale information. Last updated: 2026-08-14
Overview
Competition to unseat GPU-centric inference with custom ASICs and full rack systems optimized for serving large models.
Timeline
- 2026-08-13: Cerebras lands marquee OpenAI Ultrafast deployment vs Nvidia-centric serving (2026-08-13-openai-ultrafast-temperature2)
- 2026-07-23: Etched 10.3B (etched)
- Prior: Etched stealth/$800M era; OpenAI/Broadcom Jalapeño; MatX and others
Key Players
Analysis
Valuations are racing ahead of independent public performance proof; order books and strategic memory partners (SK Hynix) are the current credibility signals.