Cerebras CEO dismisses AI doom talk and predicts home AI appliances

AI's loudest doomsayers are talking the industry down

At TechCrunch Disrupt 2026, Cerebras Systems CEO and co-founder Andrew Feldman argued that the flood of warnings about existential AI risk is damaging the industry's public standing — and that some of the people delivering those warnings may be the wrong messengers. According to TechCrunch, he joked on stage that "some of the leaders in the AI world maybe shouldn't be allowed to talk in public."

His target was the now-mainstream discourse around dangers such as models breaking containment and insiders debating the percentage chance that AI could kill all of humanity by the end of the decade. That conversation has grown visible enough that Anthropic CEO Dario Amodei was recently parodied in an SNL sketch in which a fictional character distilled the industry's posture into one line: "We do not condone what we are doing."

Feldman's counterargument is that the upside is being drowned out. He said AI leaders with concerns should still spend more time on outcomes like a future in which your grandchildren "never meet anybody who dies of cancer."

He is not dismissive of risk. "I don't think there are powerful technologies that don't come with the challenge of 'what happens when powerful technologies are in the hands of bad people?'" he said. His position, as TechCrunch reports it, is that the technology is powerful, deserves to be handled thoughtfully, and should be regulated reasonably — a middle-ground stance from a CEO whose customers include OpenAI, Block, Figma, GSK and Lovable.

Some of the backlash is a data center problem

Feldman also offered a structural explanation for the negative mood: the data center buildout. He said the industry has spent the past few years doing a poor job of engaging local communities in the US, failing to do the community work that the power and oil-and-gas sectors have done, and that this fed the "not in my backyard" resistance now visible across the country. He contrasted the US with markets including Brazil, Canada, Finland, Iceland, Norway, Mexico and Sweden, where governments are actively encouraging data center construction.

Asked whether efficiency gains could shrink demand for his chips, he was untroubled: hardware vendors making compute more efficient and model makers making their models more efficient are complementary efforts, and meaningful progress requires all of it coming together.

Home AI appliances: 'not very far away'

In a separate Disrupt interview covered by TechCrunch, Feldman predicted that a home AI appliance is closer than most people assume. Asked what would need to happen for AI to become a household device, he answered: "Actually, I don't think that's very far away."

He pointed to advances in consumer AI agents — naming Meta's Muse and Instinct, a startup TechCrunch reports is now valued at $10 billion — as evidence that AI is finding its footing with everyday users. His specific prediction is personal AI assistants that run partly on local hardware, splitting work the way a phone does: some on device, some back in the cloud. These assistants, he said, would get to know their owners very well, sit inside a security perimeter, and "certainly have access to credit cards and be able to buy things for you."

Feldman is also a user of the technology he sells, telling the audience he now runs "a collection of agents" that sweep his calendar and pull background on people before his calls. What excites him most, he said, is not AI automating tasks people once did manually but enabling things that were previously impossible — though he is deliberately leaving the consumer-facing ideas to founders. "I like building the platforms that other people's ideas take flight on," he said.

Why it matters

Cerebras is one of the most prominent challengers to Nvidia, and Feldman's framing sketches how a major slice of the AI industry wants to be perceived amid regulatory scrutiny and public skepticism: powerful but manageable, risky but net-positive, and poorly served by its own most apocalyptic voices. His home appliance prediction also signals where chipmakers see the next demand wave — hybrid on-device and cloud inference, with assistants trusted enough to hold payment credentials. If that vision arrives on anything like his timeline, the debates over privacy, security perimeters and community acceptance of data centers are set to become more, not less, intense.