At TechCrunch Disrupt 2026, Cerebras CEO Andrew Feldman offered a two-part message: AI is about to become a household utility, and the industry's doomsayers are overcomplicating the story. The chipmaker's chief used the same interview to make both points, and together they form a distinctly pro-infrastructure argument for the technology's future.
Feldman said personal AI assistants running locally are not very far away, pointing to consumer agents such as Meta's Muse and Instinct as early proof. He described a hybrid model similar to smartphones, with some processing on-device and some in the cloud. These assistants, he argued, would know users intimately, sit behind a security perimeter, and have the ability to make purchases. He added that he already relies on a collection of agents to manage his calendar and prepare for calls.
On risk, Feldman pushed back against what he called an overly negative narrative, suggesting some AI leaders should not be allowed to speak publicly. He argued that the same technology could produce breakthroughs like eliminating cancer deaths, and that powerful tools always carry the risk of misuse. He called for thoughtful regulation rather than panic. Separately, he criticized the data center industry for failing to engage local communities in the U.S., saying that poor outreach had fueled NIMBY opposition and that other countries are more welcoming.
The two messages are not in conflict; both frame AI's challenges as manageable and its upside as underappreciated. That optimism is convenient for a company selling AI chips, but Feldman's comments also reflect a broader business argument: consumer demand and public acceptance are the real bottlenecks, not computing power.