Cerebras CEO Andrew Feldman Rejects AI Doomsday Talk, Backs Regulation
Cerebras CEO Andrew Feldman rejects AI doomsday rhetoric, urges reasonable regulation and better community engagement as data center demand keeps growing.
Summary
On October 13, 2026, Cerebras Systems CEO and co-founder Andrew Feldman argued at Disrupt 2026 that AI leaders’ doomsday rhetoric harms the industry by overshadowing potential benefits. Public debate now includes models escaping containment and estimates that AI could kill humanity by decade’s end, while Anthropic CEO Dario Amodei’s warnings were parodied on “Saturday Night Live.” Feldman said leaders should also describe outcomes such as today’s grandchildren never losing anyone to cancer.
Feldman acknowledged that powerful technology can reach bad actors and called for thoughtful, reasonable regulation. He has a commercial stake in AI adoption: Cerebras supplies OpenAI, Block, Figma, GSK and Lovable, and competes with Nvidia in AI chips.
He partly blamed opposition to US data centers on builders’ poor engagement with local communities, contrasting that resistance with encouragement from Brazil, Canada, Finland, Iceland, Norway, Mexico and Sweden. Feldman expects efficiency gains from hardware vendors and model developers to combine into greater progress, rather than reducing demand for Cerebras or rival chips.
Positives
- Cerebras supplies OpenAI, Block, Figma, GSK and Lovable, giving its chips exposure across AI, design, finance and pharmaceuticals.
- Feldman supports thoughtful, reasonable AI regulation while preserving development of potentially transformative applications.
- Brazil, Canada, Finland, Iceland, Norway, Mexico and Sweden are encouraging companies to build data centers.
- Hardware and model efficiency gains can work together to advance AI without necessarily reducing chip demand.
- Potential medical benefits could include a future in which today’s grandchildren no longer lose people to cancer.
Risks & concerns
- AI warnings now include models escaping containment and claims that the technology could kill humanity by decade’s end.
- US data center builders failed to engage local communities adequately, contributing to opposition against new construction.
- Bad actors can exploit powerful technology, creating risks that Feldman says require thoughtful regulation.
- Cerebras benefits commercially from AI adoption, giving Feldman a vested interest in countering pessimistic narratives.
- Efficiency improvements could threaten hardware demand, although Feldman rejects that outcome for Cerebras and Nvidia competitors.