OpenAI–Cerebras: 750MW wafer-scale inference pact powers GPT-5.6 Sol at 750 tok/s
Under a January 2026 Master Relationship Agreement (> $20B, 750MW of Cerebras wafer-scale capacity 2026–2028, expandable to 2GW by 2030), OpenAI serves GPT-5.6 Sol on Cerebras at up to 750 tokens/second (vs ~40–120 on GPU clusters; Cerebras
On 2026-07-01, the verified AI news record added a significant company strategy & platforms development: Under a January 2026 Master Relationship Agreement (> $20B, 750MW of Cerebras wafer-scale capacity 2026–2028, expandable to 2GW by 2030), OpenAI serves GPT-5.6 Sol on Cerebras at up to 750 tokens/second (vs ~40–120 on GPU clusters; Cerebras CEO claims up to 15x), initially for select customers from July.
Context
Under a January 2026 Master Relationship Agreement (> $20B, 750MW of Cerebras wafer-scale capacity 2026–2028, expandable to 2GW by 2030), OpenAI serves GPT-5.6 Sol on Cerebras at up to 750 tokens/second (vs ~40–120 on GPU clusters; Cerebras CEO claims up to 15x), initially for select customers from July. Verifies wide01/wide04. Strategic read: inference speed as product differentiator; OpenAI diversifying serving stack beyond GPU/Broadcom paths; Cerebras also signed AWS partnership (March 2026).
What changed
Under a January 2026 Master Relationship Agreement (> $20B, 750MW of Cerebras wafer-scale capacity 2026–2028, expandable to 2GW by 2030), OpenAI serves GPT-5.6 Sol on Cerebras at up to 750 tokens/second (vs ~40–120 on GPU clusters; Cerebras CEO claims up to 15x), initially for select customers from July. According to OpenAI/Cerebras disclosures (via Servola B, cryptobriefing B, alphatarget); VentureBeat (June 26 preview), the supporting record states: “Behind it sits a binding Master Relationship Agreement worth over 20 billion USD, which OpenAI and Cerebras have disclosed. It covers 750 megawatts of wafer-scale inference capacity from 2026 through 2028, with provisions to expand to 2 gigawatts by 2030.”.
Why it matters
Verifies wide01/wide04. Strategic read: inference speed as product differentiator; OpenAI diversifying serving stack beyond GPU/Broadcom paths; Cerebras also signed AWS partnership (March 2026). The strategy angle matters because platform control, pricing, partnerships, and access rules can shift competition even when model capabilities are similar.
Details
The research file records the item under “OpenAI–Cerebras: 750MW wafer-scale inference pact powers GPT-5.6 Sol at 750 tok/s” with source timing of July 1–13, 2026. The captured research confidence note is: High (deal disclosed; speed figures vendor-stated). Status: deal / partnership / platform move.. Additional captured source links are listed below so readers can inspect the evidence trail rather than rely on a single summary. Verifies wide01/wide04. Strategic read: inference speed as product differentiator; OpenAI diversifying serving stack beyond GPU/Broadcom paths; Cerebras also signed AWS partnership (March 2026).
Limitations and caveats
The research file did not identify a blocking caveat, but vendor-supplied claims should still be read as company statements unless independently confirmed.
Sources
Update note: Last reviewed 2026-07-22. Next checkpoint: monitor official channels and the linked source record.
Sources
Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.