TL;DR
Cerebras has secured a $10 billion compute supply deal with OpenAI, committing 750 MW of AI capacity through 2028, cementing the startup as a tier-one infrastructure partner to the world's most prominent AI lab.
What happened
- $10 billion deal signed between Cerebras and OpenAI for AI compute supply running through 2028.
- 750 megawatts of computing capacity committed, a substantial infrastructure footprint by any measure.
- OpenAI CEO Sam Altman publicly named Cerebras a "close partner," lending the deal high-visibility executive endorsement.
- The agreement positions Cerebras as a long-term, contracted supplier rather than a spot-market vendor.
Why it matters
- Cerebras gains revenue certainty at a scale that rivals the contracts underpinning Nvidia-dependent hyperscalers, validating its wafer-scale chip architecture as production-grade.
- A 750 MW commitment signals OpenAI is diversifying its compute supply chain beyond GPU-centric providers like Nvidia and Microsoft Azure.
- The deal gives Cerebras leverage for future fundraising, partnerships, and a potential IPO, having already filed for one prior to this announcement.
- For the broader market, a named $10B contract from OpenAI is a credibility signal that non-GPU architectures can compete for frontier AI workloads.
- Competitors including Groq, SambaNova, and custom silicon teams at Google and Amazon now face a better-capitalized, OpenAI-endorsed rival.
What to watch next
- Whether Cerebras expands the deal scope or MW commitment before 2028, which would confirm OpenAI is accelerating its compute diversification strategy.
- How Nvidia and Microsoft respond, given that any shift of OpenAI workloads toward Cerebras hardware reduces dependence on their infrastructure stacks.
- Cerebras IPO timing and valuation: a locked-in $10B customer contract materially changes the public-market story and could accelerate the listing.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.