Cerebras Secures $20B OpenAI Deal; Nvidia Builds 35 AI Supercomputers in Europe

A new inference architecture deal and a continental compute push: Cerebras lands its largest contract to date as Nvidia accelerates European AI research infrastructure.

🤖 Cerebras Secures $20B Multi-Year OpenAI Inference Deal and AWS Partnership

Decoded: Cerebras Systems (CBRS) announced on June 23 a multi-year agreement with OpenAI to deploy 750 megawatts of Cerebras inference compute, valued at over $20 billion — the single largest chip infrastructure contract Cerebras has disclosed since its IPO. The company simultaneously launched a parallel multi-year partnership with Amazon AWS to bring Cerebras’ high-speed inference capabilities to AWS customers. The announcements accompanied Cerebras’ Q1 2026 earnings report, which showed core revenue growth of 92% year over year. Cerebras raised $6.4 billion in Q2 2026 in what the company described as the largest semiconductor IPO in history. (Cerebras Systems official investor relations press release, GlobeNewswire, June 23, 2026)

Why it matters: A $20 billion multi-year inference deployment contract with OpenAI is a fundamental shift in Cerebras’ commercial position. Unlike Nvidia’s inference business, which is spread across thousands of enterprise buyers, Cerebras has concentrated its revenue into a small number of very large, structured agreements that carry multi-year visibility with minimal sales overhead. The AWS partnership is equally significant: it gives Cerebras distribution through the largest cloud platform in the world, moving its inference chips from a product developers must specifically seek out to one available through standard AWS procurement workflows. For investors tracking the AI inference buildout, Cerebras (CBRS) now represents the clearest public vehicle for the structural shift from training-dominated AI compute to inference-at-scale — a transition that OpenAI’s deployment choices confirm is already underway at the 750MW level.

🗄️ Nvidia Announces Record 35 AI HPC Supercomputers Across Europe at ISC 2026

Decoded: At the ISC High Performance 2026 conference in Hamburg, Nvidia announced that a record 35 AI HPC supercomputers are in development across Europe, equipping more than 3 million researchers with next-generation AI and scientific computing infrastructure. The systems are being deployed through national research programs and EU member state initiatives across the continent. Nvidia described the rollout as the largest coordinated AI research compute deployment in European history. The announcement reflects Europe’s accelerating government-backed AI infrastructure investment — a sovereign compute buildout that runs parallel to, and largely independent from, hyperscaler data center construction. (NVIDIA Newsroom, ISC High Performance 2026, June 23, 2026)

Why it matters: National HPC investment in Europe represents a structurally different demand channel than hyperscaler GPU procurement. European research computing budgets flow through sovereign funding mechanisms — the European High Performance Computing Joint Undertaking (EuroHPC JU) and national programs — which are largely insulated from the cyclical capital expenditure decisions that govern Meta, Microsoft, or Google’s GPU purchasing. A record 35 systems in active development signals that European government AI compute investment is accelerating in parallel with the private hyperscaler buildout, extending the addressable timeline for Nvidia H100 and B200 GPU demand well beyond the next hyperscaler capex cycle. For investors tracking Nvidia (NVDA) revenue diversification, the European sovereign HPC channel provides a revenue base less correlated to U.S. tech spending and more tied to multi-year national budget allocations, which are rarely reversed once committed.

Stay decoded. See you tomorrow.

— The Get AI Decoded Team