Cerebras CEO pushes back on AI doomsday warnings

Cerebras CEO Andrew Feldman has criticized the AI doomsday narrative that is currently prevalent, arguing it is not beneficial for the industry. During a presentation at TechCrunch Disrupt 2026, Feldman suggested that some prominent figures in the AI sector might need to refrain from public speaking due to their negative outlook.

The concern over AI's potential risks, such as models breaking containment or causing widespread human extinction within the next decade, has become so widespread that it has even been parodied on "Saturday Night Live," with the fictional character stating, "We do not condone what we are doing."

Feldman believes that the focus on AI's potential benefits should be emphasized more, as the technology could lead to a future where your grandchildren never meet anyone who dies of cancer. Despite his vested interest in AI's success, given his company's clients include major players like OpenAI, Block, Figma, GSK, and Lovable, Feldman remains pragmatic about the trade-offs between risk and reward.

He acknowledged the potential dangers of powerful technologies falling into the wrong hands, stating that making a technology more powerful cannot be achieved without considering the associated challenges.

Furthermore, Feldman pointed out that some of the negativity surrounding AI may be due to the data center industry's lack of engagement with local communities during the recent boom in data center construction. He criticized the industry's failure to learn from the energy sector, particularly in oil and gas, which had better practices in working with communities.

Feldman highlighted countries like Brazil, Canada, Finland, Iceland, Norway, Mexico, and Sweden, which are actively encouraging AI companies to build data centers and engage with local populations.

Lastly, Feldman expressed no concern that advancements in efficiency would lead to reduced demand for his chips or those of competitors, as he believes both hardware vendors and model makers must work together to achieve meaningful progress in AI.