Cerebras CEO to question AI scaling limits at Disrupt 2026

Andrew Feldman, CEO of AI chipmaker Cerebras Systems, will take the stage at TechCrunch Disrupt 2026 next week to ask whether the exponential growth of AI can continue unchecked. The question isn’t theoretical. It’s about whether the compute, energy, and infrastructure underpinning today’s models can scale fast enough to match the ambitions of startups and incumbents alike—or if the industry is barreling toward a bottleneck it hasn’t yet acknowledged.

Feldman’s session arrives at a moment when the AI sector’s hunger for compute has become a defining constraint. Cerebras has previously positioned itself as an alternative to traditional approaches, but even that may face limits. The broader tension isn’t just about hardware innovation; it’s about whether the ecosystem—data centers, power grids, supply chains—can keep pace with evolving demands. When Ricursive Intelligence’s founders took the Disrupt stage last week, their discussion hinted at potential workarounds. Feldman’s talk may push the conversation further: What happens if those, too, reach their limits?

The timing of this discussion is telling. Just days ago, Anthropic, Gamma, and Clay took the Disrupt stage to detail the practical challenges of deploying AI in enterprise settings. Their message was clear: The gap between demo and daily use is widening, and infrastructure is part of the reason why. Cerebras has suggested it could address some of these problems, but Feldman’s framing implies a larger concern. If AI’s compute demands continue to grow at their current pace—a trend some have observed—something may have to change. Either costs adjust, energy requirements become unsustainable, or models face new constraints.

For founders, this isn’t just a technical debate. It’s a strategic one. Startups operating under the assumption that compute will remain readily available may be making a risky bet. Investors, too, will be listening closely. When TechCrunch Disrupt’s investor sessions covered AI’s impact on VC earlier this week, the subtext was clear: The rules of the game are shifting, and not just because of hype cycles or valuations. The underlying limits of AI infrastructure could reshape which startups thrive and which struggle.

Feldman’s talk won’t offer easy answers. But it will force the audience to confront a question the AI industry has largely sidestepped: What happens when the current trajectory hits its limits? The answer will determine whether Cerebras remains a niche player or becomes a critical part of the next phase. Either way, the conversation at Disrupt 2026 just got a lot more urgent.

Sources: techcrunch.com