Cerebras CEO Andrew Feldman to Address AI Scaling
Cerebras CEO Andrew Feldman will speak at TechCrunch Disrupt 2026 on whether AI can keep scaling amid hardware and energy constraints, following the
Cerebras CEO Andrew Feldman will speak at TechCrunch Disrupt 2026 on whether AI can keep scaling. His session, titled "Can AI Keep Scaling?", will explore the growing demand for compute, energy, and infrastructure as hardware limits loom.
The event takes place October 13-15 at Moscone West in San Francisco. Feldman, who co-founded Cerebras Systems in 2015, will discuss how his company's wafer-scale computing approach tackles these constraints differently. He has described speed as a direct productivity factor in artificial intelligence.
Cerebras advances wafer-scale AI infrastructure with CS-4 launch
In August, Cerebras introduced the CS-4, the latest generation of its wafer-scale AI infrastructure. The system uses three of the company's wafer-scale engines. It delivers significant performance gains over the previous CS-3 model.
| Metric | Improvement over CS-3 |
|---|---|
| Speed | Up to twice as fast |
| Throughput per watt | Up to 10 times more |
The CS-4 was introduced on August 18. Cerebras developed wafer-scale computing by building processors on entire silicon wafers instead of cutting them into individual chips. The company was founded to challenge the assumption that powerful AI requires conventional chip architectures.
Massive data center commitments signal surging AI compute demand
Cerebras reported more than 600 megawatts of data center capacity live or under contract for delivery by the end of 2027. The company said it was increasing manufacturing capacity more than tenfold during 2026.
It plans to bring its first European data center capacity online in 2026. The company aims to expand European data center capacity to 200 megawatts by the end of 2027. Cerebras provides AI compute through on-premise systems and a cloud platform.
Strategic AI partnerships drive long-term inference workload growth
The company has secured major multiyear agreements to supply inference compute. It signed a deal with OpenAI to deploy 750 megawatts of Cerebras systems from 2026 through 2028. This contract is focused on inference workloads.
The capacity is scheduled for delivery in stages.
| Delivery Schedule | Megawatts |
|---|---|
| By end of 2026 | 250 MW |
| By end of 2027 | 250 MW |
| By end of 2028 | 250 MW |
On September 29, 2026, General Compute announced a multi-year deployment partnership with Cerebras. This deal targets agentic coding workloads. Systems from this partnership are set to come online in the first quarter of 2027.
Before founding Cerebras, Andrew Feldman co-founded and led SeaMicro, which AMD acquired in 2012. He also held leadership roles at Force10 Networks and Riverstone Networks. Feldman's appearance at TechCrunch Disrupt comes as the company, which raised $5.5 billion in a May IPO, pushes a major infrastructure expansion. Cerebras aims to expand its European data center capacity to 200 megawatts by the end of 2027.
