Andrew Feldman, CEO and co-founder of Cerebras Systems, challenged the prevailing doomsday rhetoric around artificial intelligence during his appearance at TechCrunch Disrupt 2026. Feldman criticized some AI leaders for amplifying fears about AI risks, such as the possibility of AI breaking containment or leading to humanity’s extinction by the decade’s end. He argued that this focus on potential dangers overshadows the transformative benefits AI can offer, like drastically improving human health by potentially eradicating diseases such as cancer in future generations.

Feldman emphasized that while AI is a powerful technology warranting thoughtful regulation, it inevitably carries risks—particularly if misused by bad actors. He stressed the importance of a balanced perspective that acknowledges AI’s promise alongside its challenges. His company, Cerebras, which supplies hardware to major AI clients like OpenAI, Block, GSK, and Figma, benefits from AI’s growth but maintains a pragmatic outlook on mitigating associated risks.

Addressing issues in the data center industry spurred by AI demand, Feldman pointed out that companies fell short in community engagement, especially in the U.S., which fueled local opposition to new buildouts. He compared this unfavorably to sectors like oil and gas, which collaborate more closely with communities. Feldman noted that other countries such as Brazil, Canada, Finland, and Sweden have been more welcoming to data center expansion, easing infrastructure development for AI advancements.

Looking to the future, Feldman predicted that personal AI assistants will soon become as ubiquitous as household appliances, running both on local devices and cloud infrastructure. These assistants will be highly personalized, secure, and capable of managing tasks like shopping through linked credit cards. Feldman himself uses multiple AI agents to optimize his daily workflow. He sees his role in infrastructure as enabling innovators to realize novel applications of AI, confident that improving both hardware efficiency and AI models will drive meaningful progress.