Andrew Feldman co-founded Cerebras Systems in 2016, betting that a single, wafer-sized chip could outperform clusters of smaller AI processors for certain large-model training and inference workloads. As CEO, he guided the company through a decade of private fundraising culminating in its May 2026 Nasdaq listing, one of the highest-profile AI-hardware IPOs to date.
Feldman has consistently positioned Cerebras's wafer-scale approach as a direct architectural alternative to Nvidia's GPU-cluster model, arguing that eliminating inter-chip networking overhead matters increasingly as AI models continue to grow in size.