General Compute Raises $100 Million to Build an Inference Chip Cloud

General Compute, a young company renting out artificial intelligence computing power built on specialized inference chips, has raised a $100 million Series A, TechCrunch reported on Tuesday. The round values the startup at $800 million after the money, up from $60 million following its seed round in May.

The raise is a sharp bet on a corner of the AI market that is drawing both heavy investment and growing doubts.

The Round and the Backers

According to TechCrunch, the round was led by Blackbird Ventures and Abstract Ventures. James Palmer, a principal at Blackbird, pointed to the firm's earlier investment in Baseten, an AI inference company, as part of its thinking about where demand for computing is heading.

The jump in valuation over roughly five months is unusual even by the standards of the AI boom. It suggests investors see General Compute as more than a reseller of server time.

For founders, it is also a reminder of how quickly a company's value can be rerated once it can show signed contracts, secured hardware and a credible path to power, the three things in shortest supply across AI infrastructure.

A Different Kind of Cloud

Most of the companies known as neoclouds build on Nvidia graphics processors. General Compute instead uses chips designed specifically for inference, the work of running trained AI models to answer requests, from makers including SambaNova, Positron and Cerebras.

The company has agreed a $400 million chip purchase deal with SambaNova, which TechCrunch described as the first debt financing of inference chips. It has also signed a large customer contract worth $100 million in annual recurring revenue.

Rather than waiting for giant new campuses, General Compute uses smaller power sources, rents space inside other data centers and converts former bitcoin mining sites. The company expects to have 75 megawatts deployed in 2027 and access to 200 megawatts of power.

Power has become one of the tightest constraints in the AI build out. By looking for capacity in places larger players overlook, the company is trying to grow faster than it could if it had to wait years for new grid connections.

The Founder's Job Now

Co-founder and CEO Finn Puklowski told TechCrunch that raising money will dominate his time. "Capital is going to be a full time job for me, both on debt and equity," he said.

That is a frank description of the business. Building computing capacity requires buying chips and securing power long before customers pay, so the founders of infrastructure companies spend much of their time lining up lenders as well as investors.

Debt is attractive because it lets a company grow without giving away more ownership, but it also adds pressure. Loans backed by chips must be repaid whether or not the hardware stays busy, which makes steady customer demand essential.

The Doubts in the Market

The raise lands at a nervous moment for the sector. TechCrunch noted that shares in CoreWeave, the best known neocloud, were trading near a 12 month low, and that Elon Musk has predicted that "200+ neoclouds are going to see guppy level death and consolidation."

Puklowski said such fears are "massively overstated." SambaNova CEO Rodrigo Liang framed the real test more bluntly: "Everyone in this market is selling speed. The real question is whether you can make money doing it."

Lessons for Founders

General Compute's story offers a few lessons. It picked a clear point of difference in its choice of chips rather than competing head on with the largest players. It found creative ways around the power shortage that slows many rivals. And it secured a large customer commitment that gives investors evidence of demand.

The hard part comes next. With $100 million raised and a $400 million chip deal signed, the company now has to turn hardware and megawatts into profitable revenue in a market that is crowded and unforgiving.