Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
IndustryInsight · On RadarTechCrunch→ original

General Compute Raises $100M at $800M Valuation, Buys Chips From Someone Other Than Nvidia

General Compute has raised $100M in a Series A round at an $800M valuation, up from $60M after its May round. The company rents out servers built on chips from SambaNova, Cerebras, and Positron rather than Nvidia GPUs, so customers can run ready-made models. The money arrived as CoreWeave shares hit a 12-month low. If you run open models, it's worth comparing price and speed on non-Nvidia chips.

AI-processed from TechCrunch; edited by Hamidun News
General Compute Raises $100M at $800M Valuation, Buys Chips From Someone Other Than Nvidia
Source: TechCrunch. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceTechCrunch ↗
Editor-in-Chief
$400Mthe deal for SambaNova chips, the first debt financing for inference chips
Key metric #2A customer with $100M in annual recurring revenue opened the door to the round
2027 plan75 MW of capacity, with access to 200 MW

General Compute has raised $100M in a Series A round, the first major round after its initial funding. Investors valued the company at $800M post-money. After its May round, it was worth $60M. General Compute is building a "neocloud," the term for companies that rent out computing power for AI. Blackbird Ventures and Abstract Ventures led the round. It was announced in a bad week for the sector.

Background: neoclouds under pressure

CoreWeave, a public neocloud, has fallen to a 12-month low. Elon Musk, meanwhile, boasts of $40B in debt to buy compute and predicts that "200+ neoclouds face guppy-level death and consolidation." General Compute CEO Finn Pookloski disagrees. In his view, the fears of a crash are "greatly exaggerated": he gets calls every week asking him to find any chips at all, whether TPUs, GPUs, or something else. He thinks Musk is acting in his own interest and depressing GPU prices that Musk himself wants to buy. Pookloski says Anthropic's optimistic S-1 filing "removes fears for today." This is the opinion of a founder who is raising money himself, and it should be weighed accordingly.

How General Compute differs from the rest

While nearly everyone chases GPUs, the company is assembling a cloud from chips designed for inference, meaning running already-trained models. Nvidia's competitors make such chips. General Compute has struck a $400M deal to buy new chips from SambaNova. According to TechCrunch, this is the first debt financing for inference chips. Positron and Cerebras have now joined its suppliers.

Its customers are not ordinary users but cloud services that sell fast inference to developers. General Compute provides bare metal: bare servers with no software layered on top. The customer installs everything else and brings its own product to market. The path to the round was opened by a contract with a major client worth $100M in annual recurring revenue.

The second difference concerns sites. The company is not building huge data centers from scratch. It uses small power sources, leases space in other companies' data centers, and converts former Bitcoin mining facilities for running AI models. Pookloski explains it this way: when the company has chips, it will have somewhere to plug them in.

This approach has a cost. General Compute will not enter the top tier of neoclouds or compete with hyperscalers, the largest clouds such as AWS, anytime soon. For 2027, the company plans to deploy 75 megawatts of compute capacity. Access to electricity could reach 200 MW, but according to Pookloski, that is "more than we can likely finance and supply with hardware." Money is not fully settled either: "capital will be a full-time job for me, both debt and equity."

What investors and chipmakers say

James Palmer of Blackbird, who co-led the round, points to experience with Baseten, an inference cloud. The fund has watched its growth over the past 10 months. He sees demand for inference of open models and companies' desire to own their own intelligence. In his words, Cerebras, Positron, and SambaNova need distribution partners who can handle the difficulties of the physical world: electricity, sites, and supply. Executives at all three chipmakers praised General Compute for focusing on the economics of cloud computing. SambaNova head Rodrigo Liang said: "Everyone in this market sells speed. The real question is whether you can make money on it."

What this means for practitioners

If you run open models in production, you have new options. Inference is becoming a separate market with its own hardware, and non-Nvidia chips are winning credit and major customers. It makes sense to compare providers' price and speed on such chips, not only on GPUs.

There are limits too. General Compute's capacity is still measured in tens of megawatts. Its future depends on whether it can secure debt and the chips themselves. The debate over the fate of neoclouds is not settled: CoreWeave's fall and General Compute's success together show that investors pay those who have a clear way to make money.

Story

Key Questions Answered

What are the key details on Background: neoclouds under pressure?

CoreWeave, a public neocloud, has fallen to a 12-month low. Elon Musk, meanwhile, boasts of $40B in debt to buy compute and predicts that "200+ neoclouds face guppy-level death and consolidation." General Compute CEO Finn Pookloski disagrees. In his view, the fears of a crash are "greatly exaggerated": he gets calls every week asking him to find any chips at all, whether TPUs, GPUs, or something else. He thinks Musk is acting in his own interest and depressing GPU prices that Musk himself wants to buy.

How General Compute differs from the rest?

While nearly everyone chases GPUs, the company is assembling a cloud from chips designed for inference, meaning running already-trained models. Nvidia's competitors make such chips. General Compute has struck a $400M deal to buy new chips from SambaNova. According to TechCrunch, this is the first debt financing for inference chips. Positron and Cerebras have now joined its suppliers. Its customers are not ordinary users but cloud services that sell fast inference to developers.

What investors and chipmakers say?

James Palmer of Blackbird, who co-led the round, points to experience with Baseten, an inference cloud. The fund has watched its growth over the past 10 months. He sees demand for inference of open models and companies' desire to own their own intelligence. In his words, Cerebras, Positron, and SambaNova need distribution partners who can handle the difficulties of the physical world: electricity, sites, and supply.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Private AI Builders Community · Vibe Club

Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews

Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.

Join Vibe Club →Explore Hamidun Academy
What do you think?
Loading comments…