Fares Hassanieh

Last month, @cerebras IPO'd in the largest semiconductor IPO in history. In 2016, we started with a vision, a little bravado, and two contrarian bets. We saw AI on the horizon. And for computer architects, big new workloads are exciting opportunities. Absent a new workload, the status quo dominates. For nearly 30 years the server chip market was a two company game. @intel & @AMD's x86 chips had powered every PC and laptop and most servers for generations. It was nearly impossible to break in. But when mobile computing emerged, neither Intel nor AMD won any share. @Arm won it all. A new workload. A new winner. 15 years later, the cell phone processor market is 100% ARM. The same had been true for graphics workloads. The obvious answer would have been for existing x86 processor companies to win. But a new player -- Nvidia won -- with a new part. A part optimized exclusively for graphics. A Graphics Processing Unit, the GPU. With history as our guide, the Cerebras founders bet that AI would be next. That it would require a new architecture. A dedicated one. From the ground up, optimized in every way for AI work. So we set out to build it. Not a modified GPU. A clean sheet design. Designed for AI and only AI. Our bet was 100% contrarian. It turned out to be dead right. Every major shift in computing creates opportunity. Catching the opportunity requires some vision, bravado, the will to build something new, more time than you planned and a bit of luck.