Gimlet Labs

Builds a multisilicon inference cloud that routes AI workloads across heterogeneous hardware, helping frontier labs and other large-scale inference users improve speed, throughput, and power efficiency.