Cerebras CEO Rejects AI Doomsday Rhetoric
Cerebras Systems CEO Andrew Feldman criticized AI leaders for pushing existential threat narratives, arguing the focus should shift to the transformative benefits of the hardware and models.
Speaking at TechCrunch Disrupt 2026, Cerebras Systems chief executive Andrew Feldman pushed back sharply against the apocalyptic warnings currently dominating public discussions about artificial intelligence. Feldman suggested that some AI executives "maybe shouldn't be allowed to talk in public" due to their focus on doomsday scenarios. He noted that widespread discourse surrounding models breaking containment or predictions of human extinction by the end of the decade—even parodied on Saturday Night Live with a fictional portrayal of Anthropic CEO Dario Amodei—has created an unnecessarily negative perception of the sector.
Rather than focusing entirely on existential risks, Feldman argued that leaders should highlight the immense potential of AI. He stressed that while powerful technologies inherently carry risks if placed in the hands of bad actors, society must balance safety with optimism. For instance, he noted that AI advancements could create a future where coming generations never see anyone die of cancer. To achieve this safely, he advocated for thoughtful, reasonable regulation rather than panic.
Feldman also attributed growing public skepticism to how the data center industry has handled infrastructure buildouts. He observed that AI infrastructure builders failed to engage local U.S. communities as effectively as oil, gas, and power industries historically have, fueling domestic opposition. In contrast, foreign markets including Brazil, Canada, Finland, Iceland, Norway, Mexico, and Sweden have taken a more welcoming approach to data center projects.
Despite ongoing efforts to optimize both hardware and software, Feldman remained unconcerned that efficiency improvements would decrease chip demand for Cerebras or rival Nvidia. Instead, he maintained that hardware efficiency and model-level progress must advance simultaneously. Cerebras continues to sell its compute technology to a diverse group of high-profile clients, including OpenAI, Block, Figma, Lovable, and global pharmaceutical company GSK.
This is our own summary of reporting by TechCrunch AI
