General Compute buys NVIDIA GPUs and specialized inference chips from Cerebras, SambaNova, and Positron AI, deploys them in colocation, and sells the capacity to AI companies and to other inference clouds. Prefill runs on the GPUs. Decode runs on the alternative chips. Customers can take dedicated bare metal or a guaranteed tokens-per-second contract, while General Compute operates the cooling, networking, power, and the software that routes each request. The San Francisco company announced a $100 million Series A on October 27, 2026, co-led by Blackbird Ventures and Abstract Ventures, with Fuse Venture Partners, Village Global, and others participating. First major deployments are scheduled for the first quarter of 2027, with more than 50 megawatts of capacity in the pipeline. The round follows a $400 million debt facility from Upper90 for chip purchases. TechCrunch reported an $800 million post-money valuation, up from $60 million after a May seed round, and said a customer contract worth $100 million in annual recurring revenue helped open the raise. CEO Finn Puklowski and CTO Jason Goodison founded the company. It places hardware in rented data center space and converted former bitcoin-mining sites, and its early buyers are developer-facing inference clouds that run their own software on the bare metal. General Compute also offers an OpenAI-compatible API. On its site, the company claims up to 16.2 times faster end-to-end results than a GPU cloud in its own benchmarks, and shows a comparison at 2,000 tokens per second per user against 110 on a GPU cloud.