General Compute raises $100M Series A for non-GPU inference

General Compute raised a $100 million Series A at about an $800 million post-money valuation, according to TechCrunch. Blackbird Ventures and Abstract Ventures led the round, TechCrunch reports.

The company builds cloud capacity around inference chips made by NVIDIA's rivals instead of GPUs. It raised the money as investors question whether "neocloud" companies, which rent AI computing, can make a profit.

What General Compute does

General Compute sells bare-metal servers, meaning customers get whole machines rather than slices, mostly to companies that run inference clouds. TechCrunch says it uses smaller power sources, rents space in other companies' data centers and converts former bitcoin-mining sites.

It was valued at $60 million after a seed round in May, so the valuation rose more than tenfold in about five months, per TechCrunch.

The chip deals

TechCrunch reports a $400 million deal with SambaNova for new chips, described as the first debt financing of inference chips. The company has also added Positron and Cerebras as suppliers. A deal with one large customer worth $100 million in annual recurring revenue helped make the raise possible, TechCrunch says.

Plans and caveats

General Compute expects to deploy 75 megawatts of computing capacity in 2027 and to have access to 200 megawatts of power, TechCrunch reports. CEO Finn Puklowski said "capital is going to be a full-time job for me, both on debt and equity."

This article rests on TechCrunch's report. General Compute's own announcement was not available to us. Watch whether the 75 megawatts arrive on schedule in 2027.

Why it matters

Most AI cloud capacity today runs on NVIDIA GPUs. General Compute is betting that chips built only for inference, the work of running a trained model, can be cheaper to operate at scale. SambaNova's chief executive Rodrigo Liang framed the test in TechCrunch's report: "The real question is whether you can make money doing it."