General Compute Inference
Serve model predictions on specialized compute hardware in the cloud.
General Compute Funding History
General Compute rents cloud capacity for AI inference on specialized chips from suppliers other than Nvidia. Customers can rent bare-metal machines or buy capacity under agreements measured in tokens per second.
Company Overview
General Compute
Founded 2025 · Covina, California, US (per LA Business Journal; SF also reported)
Official WebsiteGeneral Compute Funding Timeline
Series A
US$100 million
General Compute said its first major deployments were due in early 2027. Its cloud runs inference on GPUs and specialized chips from Cerebras, SambaNova and Positron AI.
Led by Blackbird Ventures, Abstract Ventures.
Source · techcrunch.com ↗
General Compute Founding Team
Finn Puklowski co-founded General Compute and took on the CEO role. The company focuses on inference capacity built around alternatives to Nvidia GPUs.
- Jason Goodison — CTO. University of Waterloo graduate.
General Compute Products
General Compute Company Development
Adding a Second Accelerator Architecture
General Compute built its inference cloud around specialized accelerators, including SambaNova hardware. A September 2026 agreement added Cerebras wafer-scale systems to its deployment plans, with service scheduled for the first quarter of 2027. The expansion gives the company another hardware architecture for serving models.
General Compute Competitive Landscape
Choosing the Accelerator Beneath Inference
General Compute and Cerebras both offer compute outside Nvidia’s GPU ecosystem. General Compute supplies capacity from colocated non-Nvidia ASICs with bare-metal or throughput-based arrangements. Cerebras supplies infrastructure based on its own wafer-scale architecture.
