Cerebras Systems, a semiconductor startup, was founded in 2016 on the radical idea of making chips dramatically bigger rather than smaller, building the world’s largest computer chip that occupies an entire silicon wafer. CEO Andrew Feldman, who previously co-founded SeaMicro in 2007 with chip architect Gary Lauterbach, sold that company to AMD in 2012. After leaving AMD, Feldman began questioning the use of Nvidia GPUs for AI, viewing them as an accident of history since they were originally built for video games, not neural networks. In 2014, deep learning was still largely confined to research labs, and many experts saw AI as a niche project. The team believed AI would outgrow GPU architecture and imagined an entirely new computer designed specifically for AI. The company’s journey involved bold engineering, repeated failures, and patient investors. Today, as AI models grow into trillions of parameters, Cerebras has emerged as one of Nvidia’s most serious challengers.