General Compute Inc.

General Compute is the deployment arm for heterogeneous AI compute, helping inference clouds serve tokens up to 40x faster at a lower blended cost than GPU-only infrastructure.

General Compute pairs Nvidia GPUs for prefill with decode silicon from partners including Cerebras, SambaNova, d-Matrix and Positron, delivered under multi-year take-or-pay contracts. The company is signing $190M in annual contracted revenue with major inference clouds, and Blackbird co-led its Series A.