Fetching from the wire…
01
02
03
04
05
06
07
08
Infra2026-08-20 · source-backed
WSE-3 Turbo wafer, 4 trillion transistors, 250 PFLOPS, three processors per Nexus rack-scale system, wafer-to-wafer interconnect latency down to 2 microseconds. Claims: up to 30x faster inference than production GPU systems, 10x throughput per watt over CS-3, and 1,000+ tokens per second on models above 10 trillion parameters. First shipments this quarter, no published pricing. (Cerebras) The HN thread ran 458 points and 269 comments arguing about whether the benchmarks survive real serving workloads, which is the correct argument to have.
Each link below shares sources, entities, or timing with this story.
Cerebras partners with OpenAI / Shared entity: GPU / Earlier coverage / Tension
Linked by a graph relationship (Cerebras partners with OpenAI); both cover GPU; earlier GPU coverage from 2026-07-25.
Cerebras partners with OpenAI / Shared entity: Cerebras / Same source domain / Shared topic / Earlier coverage
Linked by a graph relationship (Cerebras partners with OpenAI); both cover Cerebras; reported by the same outlet (cerebras.ai).
Cerebras partners with OpenAI / Shared entity: GPU / Earlier coverage
Linked by a graph relationship (Cerebras partners with OpenAI); both cover GPU; earlier GPU coverage from 2026-08-05.
Cerebras partners with OpenAI / Shared entity: Cerebras / Shared topic / Earlier coverage
Linked by a graph relationship (Cerebras partners with OpenAI); both cover Cerebras; overlapping topics (cerebra, down).
Cerebras partners with OpenAI / Shared entity: Nexus / Earlier coverage / Tension
Linked by a graph relationship (Cerebras partners with OpenAI); both cover Nexus; earlier Nexus coverage from 2026-07-23.
Cerebras partners with OpenAI / Shared entity: GPU / Earlier coverage / Tension
Linked by a graph relationship (Cerebras partners with OpenAI); both cover GPU; earlier GPU coverage from 2026-06-13.
Linked by a graph relationship (Cerebras partners with OpenAI); both cover GPU; earlier GPU coverage from 2026-04-30.
Cerebras partners with OpenAI / Shared entity: GPU / Earlier coverage
Linked by a graph relationship (Cerebras partners with OpenAI); both cover GPU; earlier GPU coverage from 2026-08-08.