CoreWeave, the publicly traded neocloud that rents out AI computing power, has watched its stock tumble to a 12-month low. Meanwhile, Elon Musk is wielding $40 billion in debt to secure computing capacity and warning that "200+ neoclouds are going to see guppy-level death and consolidation." All things considered, it's hardly the perfect moment to announce a Series A for a neocloud.

General Compute co-founder and CEO Finn Puklowski isn't rattled. He remains optimistic about CoreWeave's prospects and believes Musk is simply talking his book to push down prices on the GPUs he wants to purchase. According to Puklowski, fears of a chip-driven collapse are "massively overstated. I have people calling me up every single week asking me to do TPUs, GPUs, any chip, to get them more compute." Equally significant, he notes, is that Anthropic's bullish S-1 is "settling fears for today."

General Compute has carved out a niche by constructing its cloud from specialized inference chips manufactured by Nvidia's competitors. The company secured a $400 million deal to buy new chips from SambaNova—marking the first debt financing of inference chips—and has since added Positron and Cerebras to its roster of chip suppliers. A contract with a major customer worth $100 million in annual recurring revenue opened the door for the company to raise $100 million in Series A funding led by Blackbird Ventures and Abstract Ventures. After being valued at $60 million following a May seed round, General Compute now carries an $800 million post-money valuation.

The startup's approach is deliberately contrarian. It concentrated on chips designed for inference while competitors pursued GPUs. Rather than undertaking the complex task of building massive data centers, it relies on smaller power sources, leases space in other companies' data centers, and repurposes former bitcoin-mining sites into facilities for running AI models. This strategy means General Compute won't be vaulting into the top tier of neoclouds or challenging hyperscalers anytime soon—but it also avoids the constraints that come with building data centers from the ground up. When General Compute obtains chips, it will have somewhere to plug them in. Puklowski anticipates deploying 75 megawatts of computing capacity in 2027, and he expects to have access to 200 megawatts of power—"more than we can probably fund and get hardware for."

Even so, the plan has its limits. The startup has cultivated key customers among cloud companies that specialize in providing developers with high-speed inference; General Compute supplies the bare metal—raw servers with no software layered on top—for them to bring their offerings to market. James Palmer, a principal at Blackbird who co-led the round, said his firm's confidence in General Compute stems in part from an investment in Baseten, an inference cloud company. Regarding General Compute, he said, "Cerebras, Positron, SambaNova, all need distribution partners who can take them to scale, who can navigate the complexity of the physical world." Rodrigo Liang, SambaNova's CEO and co-founder, put it this way: "Everyone in this market is selling speed. The real question is whether you can make money doing it."