General Compute raises $100M Series A at $800M valuation despite neocloud funding fears

Fuse Venture Partners, Village Global and other investors also participated. The round stood out because General Compute’s rivals in the neocloud segment—cloud providers built around AI infrastructure—are under pressure. CoreWeave Inc., the largest neocloud, recently saw its stock fall to a 12-month low amid talk that investors are weary of financing an endless, multibillion-dollar AI buildout.

Elon Musk added to those concerns last weekend in a post on X. He said SpaceX had secured $40 billion in new debt, but warned that he could get the money only because the company has alternative revenue streams. “200+ neoclouds are going to see guppy-level death and consolidation,” Musk said. “Some will make it through, but not many.”

Musk’s comments followed a forecast from Gartner Inc. that up to 80% of all neoclouds could disappear by the end of the decade. Smaller neoclouds are already feeling pressure. Earlier this month, Microsoft Corp. cancelled a multibillion-dollar deal with Nscale Global Holdings Ltd., a move that calls into question the startup’s plans for an initial public offering.

General Compute co-founder and CEO Finn Puklowski told TechCrunch he is not worried about what other companies are doing. He said Musk’s comments are likely an effort to drive down prices for the graphics processing units he wants to buy, and called fears of a chip industry collapse “massively overstated” based on the demand General Compute is seeing.

General Compute’s business model differs from that of most neoclouds. It does not focus on general-purpose GPUs; it buys inference chips from other chipmakers, including Cerebras Systems Inc., SambaNova Systems Inc. and Positron AI Inc. Customers do not manage the hardware. They set up a workload and press start, and General Compute routes it to the most suitable chip. Customers can choose dedicated bare-metal servers or a guaranteed tokens-per-second service-level agreement.

The startup also does not plan to build its own data centers. It rents space in facilities owned by other companies, which handle networking, power delivery, cooling and maintenance. In addition to the equity round, General Compute secured $400 million in debt financing in July, and most of that money has not been spent.

Puklowski said the company has been running some workloads for customers, but its first large-scale chip deployments will not go live until the first quarter of next year. It has more than 75 megawatts of capacity planned for next year and has already signed one large deal expected to bring in more than $100 million in annual recurring revenue.

Blackbird’s James Palmer said silicon is no longer the major bottleneck in the AI industry; companies need power. “Both power and silicon supply chains are slow to scale,” he said. “Finding creative ways through those constraints will decide who wins. Finn and his team are the ultimate coalition builders, and they think in decades, not days.”

Cerebras Chief Technology Officer Sean Lie said General Compute understands that inference performance depends on how the full system is deployed and operated. “We value their approach and commitment to fast execution at scale,” Lie said.