Cerebras Systems CEO Andrew Feldman spoke at TechCrunch Disrupt 2026 about the near future where personal AI assistants will become common household appliances. Feldman envisions AI agents running partly on local devices and partly in the cloud, allowing them to learn user preferences intimately while maintaining security. These assistants will handle tasks like managing credit card transactions and improving daily life, much like current consumer AI products from companies like Meta and the startup Instinct.

Feldman himself is an active user of AI agents, employing a network of them to manage his calendar and provide information before meetings. He expressed enthusiasm not only about AI performing existing tasks but also about the new possibilities the technology will unlock, though he leaves the exploration of these ideas to other innovators. As a chipmaker, Cerebras is focused on creating the infrastructure needed for these AI capabilities to flourish.

While discussing AI advancements, Feldman also addressed concerns around AI risks and the negative media narrative often dominating the conversation. He called for a more balanced discussion that highlights AI’s potential benefits, such as potentially eradicating diseases like cancer in future generations. Feldman emphasized that powerful technologies inevitably come with risks, but responsible regulation and thoughtful use are key to maximizing the technology’s positive impact.

Feldman also commented on challenges faced by the data center industry concerning AI growth, noting that the industry historically failed to engage local communities effectively in the U.S. He contrasted this with more cooperative approaches seen in countries like Brazil, Canada, and Finland. Despite efficiency gains in hardware and software, Feldman remains confident that demand for advanced AI chips will continue to grow alongside the development of smarter, more efficient AI models.