Updated
Updated · TechCrunch · Sep 30
Cerebras CEO to Discuss AI Scaling at TechCrunch Disrupt as 750-MW OpenAI Deal Expands
Updated
Updated · TechCrunch · Sep 30

Cerebras CEO to Discuss AI Scaling at TechCrunch Disrupt as 750-MW OpenAI Deal Expands

3 articles · Updated · TechCrunch · Sep 30

Summary

  • Andrew Feldman will take the TechCrunch Disrupt 2026 stage to examine whether AI can keep scaling as compute, power and infrastructure demands rise.
  • Cerebras will use the session to argue its wafer-scale architecture offers an alternative to conventional chips if current AI hardware approaches practical limits.
  • The company is already scaling around that thesis: it signed a multiyear deal with OpenAI to deploy 750 megawatts of systems from 2026 through 2028.
  • Cerebras said in August it had more than 600 megawatts of data-center capacity live or under contract by end-2027, planned a 10-fold manufacturing increase in 2026, and aims for 200 megawatts in Europe by end-2027.
  • The talk is part of TechCrunch Disrupt, set for October 13-15 in San Francisco, where more than 10,000 attendees, 250-plus speakers and 300-plus startups are expected.

Insights

Can Cerebras’ wafer-scale AI systems really outrun power, cooling, and data-center bottlenecks as demand for AI infrastructure explodes?
Will Andrew Feldman’s Disrupt 2026 vision prove that AI’s future belongs to full-stack infrastructure, not just faster processors?
If AI scaling is now limited more by memory, energy, and facilities than chips, what happens to today’s GPU-first playbook?