General Compute has raised $100 million in Series A funding as it builds an AI cloud around inference chips from Nvidia rivals. The company’s approach pairs those chips with rented data center space and former bitcoin-mining sites, rather than relying on a large buildout of new data centers.
The funding round follows a $100 million annual recurring revenue deal with a large customer. It places General Compute in a market where questions about neoclouds, chip supply and the cost of serving AI workloads have become harder to ignore.
A different bet on AI computing
Many neocloud companies have centered their businesses on GPUs. General Compute instead focused on specialized chips designed for inference, the stage where trained AI models respond to requests. Its suppliers include SambaNova, Positron and Cerebras.
The company has struck a $400 million deal to purchase new chips from SambaNova, described in the source as the first debt financing of inference chips. Adding Positron and Cerebras gives General Compute more than one chip supplier as it builds out its infrastructure.
That supplier mix reflects a broader business role: General Compute provides computing capacity to cloud companies that sell high-speed inference to developers. It supplies bare metal, meaning raw servers without software layered on top, for those customers to use in their own offerings.
Capacity without building everything from scratch
General Compute’s infrastructure strategy also differs from the effort to construct large data centers. It uses smaller power sources, rents room in other companies’ data centers and converts former bitcoin-mining locations to run AI models.
This can give the company places to install chips without taking on the same constraints involved in building data centers from scratch. It does not mean General Compute is positioned to quickly match the largest neoclouds or compete with hyperscalers. The approach is more focused: secure chips and power, then put them to work for customers.
Co-founder and CEO Finn Puklowski expects the company to deploy 75 megawatts of computing capacity in 2027. He suspects he will have access to 200 megawatts of power, which he said is more than the company can probably fund and equip with hardware.
Funding and demand shape the next steps
The Series A was led by Blackbird Ventures and Abstract Ventures. General Compute was valued at $60 million after a May seed round and is now valued at $800 million post-money to its investors.
Investor interest is tied in part to demand for inference capacity. James Palmer, a principal at Blackbird who co-led the round, said the firm had seen the growth of Baseten, an inference cloud company in which it had invested. He pointed to demand for open weight inference and for companies owning their own intelligence.
Palmer also described a need for distribution partners for Cerebras, Positron and SambaNova: companies able to help those chipmakers scale and navigate the practical complexity of physical infrastructure. General Compute is aiming to fill that role by connecting chip suppliers, computing sites and cloud customers.
Optimism amid neocloud concerns
The funding arrives amid negative signals for neoclouds. CoreWeave’s stock price has fallen to a 12-month low, while Elon Musk has pointed to $40 billion in debt for buying computing capacity and predicted that “200+ neoclouds are going to see guppy-level death and consolidation.”
Puklowski is bullish on CoreWeave’s chances and believes the warnings about a chip-led collapse are overstated. He said people call him every week seeking more compute, across TPUs, GPUs and other chips. He also sees Anthropic’s bullish S-1 as “settling fears for today.”
Demand alone does not settle the economics. SambaNova CEO and co-founder Rodrigo Liang said that while companies are selling speed, the key question is whether they can make money doing it. General Compute’s model will depend on securing capital as well as chips, power and customers. Puklowski told TechCrunch that financing through debt and equity will remain a full-time job for him.