Why This Chip CEO Says AI Won't End the World

Cerebras Systems CEO and co-founder Andrew Feldman rejected warnings that AI will bring disaster at TechCrunch Disrupt 2026. Speaking in a session titled 'Can AI Keep Scaling?', he said some AI leaders "maybe shouldn't be allowed to talk in public" about doom. TechCrunch
He described a very different future. Feldman said AI "could lead to a world where your grandkids will never meet anybody who dies of cancer." He was talking about using large AI models to study biology and improve care. TechCrunch
He did not say companies should build without rules. He said "powerful technologies come with the challenge of what happens when they are in the hands of bad people." He added that AI "is powerful, must be approached thoughtfully and regulated reasonably." TechCrunch
The business story is about inference. That means using a trained model to answer live questions. Cerebras lists customers including OpenAI, Block, Figma, GSK and Lovable. That mix covers AI chat, money tools, design tools, drug research and tools that write software. In all five, fast answers matter.
The company says its CS-4, a full rack that works as one large machine, gives up to 30x faster inference than GPUs. OpenAI's GPT-5.6 Sol runs in Ultrafast mode on Cerebras at up to 750 tokens per second. Tokens are small bits of words, so this is like typing speed for AI. Cerebras has partnered with Lovable to power faster AI software creation and with AMD on split-apart inference, which keeps computing, memory and connections separate, combining AMD Helios and the Cerebras Wafer-Scale Engine, a very large chip. Cerebras
Feldman also talked about where to build and how to power data centers. He said "the industry was not good at working with local communities in the U.S. on AI-driven data center buildouts." He named Brazil, Canada, Finland, Iceland, Norway, Mexico and Sweden as places that invite companies to build.
TechCrunch Disrupt 2026 had six stages picked by editors. Feldman's scaling session was part of the business and infrastructure program announced on Sept. 30.
The broader context here is that two questions have joined into one. Who gets to use strong AI, and under what safety rules. And where to place data centers, how to power them, and how to keep local trust. Both now shape product plans as much as chips or memory.
In my view, hope about cures and care about harm can live side by side. We have seen this before with the PC, the commercial internet, mobile phones and cloud computing. Each wave brought good uses and bad uses, and teams learned how to manage them. Now the pressure is less on training and more on answering fast, at low cost, for many users at once. Big rack systems, split-apart designs, and building where permits and power are clear are responses to that shift. Local concerns are practical, like noise, water, grid waits and power bills. They need early honesty and long-term presence. Places that give clear rules will get more builds, which can speed up useful products and medical research.


