Cerebras Systems’ Andrew Feldman on whether AI can keep scaling at TechCrunch Disrupt 2026

The cost of AI compute infrastructure is falling as wafer-scale alternatives emerge.

Cerebras Systems CEO Andrew Feldman will discuss the future of AI scaling at TechCrunch Disrupt 2026. The company challenges conventional AI hardware by focusing on wafer-scale computing, where an entire silicon wafer is used as a single processor rather than being cut into smaller chips. Cerebras has developed AI compute solutions available through on-premise systems and its cloud platform. Feldman's session will address the increasing demand for compute, energy, and infrastructure required for AI advancements. He will also explore Cerebras' alternative approach to these constraints and potential outcomes if current AI hardware reaches its limitations. Cerebras recently raised $5.5 billion in an IPO and secured a multiyear deal with OpenAI. The company also introduced its latest wafer-scale AI infrastructure, CS-4, and is expanding its data center capacity globally.

Cerebras' focus on wafer-scale computing offers a direct challenge to the economics of AI infrastructure, potentially lowering the barrier to entry for advanced AI development. By leveraging an entire wafer as a single processor, the company aims to overcome the physical and energetic limitations of traditional chip architectures. This approach could significantly alter the landscape for AI hardware providers and the companies that rely on their services, particularly in managing the escalating demands for compute power.

The broader implications extend to the scaling of AI itself. Feldman's discussion at Disrupt 2026 is poised to shed light on whether current hardware paradigms can sustain AI's rapid evolution or if alternative solutions like wafer-scale computing are essential. The success and expansion of Cerebras' business, including its significant deal with OpenAI and its growing data center capacity, suggest a market readiness for such innovations. This signals a potential shift in how AI compute is provisioned and consumed, impacting everything from cloud providers to enterprise AI deployments.