General Compute, a San Francisco company that runs cloud capacity for AI inference, has raised $100m in a Series A round co-led by Blackbird Ventures and Abstract Ventures. The company announced the round on 27 October 2026. Fuse Venture Partners and Village Global also took part, along with investors it did not name.
The round values General Compute at $800m after the investment, according to TechCrunch. The publication reported that a seed round in May 2026 had valued the company at $60m post-money.
TechCrunch said a deal with one large customer, worth $100m in annual recurring revenue, helped pave the way for the raise. The customer was not named.
“I have people calling me up every single week asking me to do TPUs, GPUs, any chip, to get them more compute.”Finn Puklowski, Co-founder and CEO, General Compute (to TechCrunch)
What General Compute does
General Compute calls itself an inference cloud for very fast token generation. It finances, installs and operates GPUs and chips designed specifically for running AI models, and assigns each customer's workload to the hardware that suits it best. Its chip partners are Cerebras, SambaNova and Positron AI.
Customers can rent dedicated bare-metal servers or buy a service level that guarantees a set number of tokens per second, without having to buy or manage the hardware themselves. TechCrunch reported that its main customers so far are cloud companies that sell high-speed inference to developers and use General Compute's raw servers underneath their own offerings.
Rather than building large data centres, the company uses smaller power sources, rents space in other operators' facilities and converts former bitcoin-mining sites, according to TechCrunch. Crypto Briefing reported that the company was founded in 2025 and is led by chief executive Finn Puklowski and chief technology officer Jason Goodison.
Debt for chips, equity for growth
The equity round sits on top of a debt facility of up to $400m from Upper90 that General Compute had already announced. Crypto Briefing reported that the facility was agreed in July 2026, that its first draw was $100m and that the company's inference chips serve as part of the collateral. TechCrunch described the company's $400m chip purchase from SambaNova as the first debt financing of inference chips.
Lenders have usually preferred Nvidia hardware as collateral, which makes the arrangement an early test of whether rival chips can be financed in the same way.
Puklowski told TechCrunch that raising money would remain a full-time job for him on both the debt and equity sides.
Capacity plans
General Compute says its first major deployments go live in the first quarter of 2027 and that it has more than 50 megawatts of capacity in its pipeline. Puklowski told TechCrunch he expects to deploy 75 megawatts in 2027 and may have access to 200 megawatts of power, more than the company can likely fund and fill with hardware.
James Palmer, a principal at Blackbird, told TechCrunch that chipmakers such as Cerebras, Positron and SambaNova need distribution partners that can take them to scale and handle the physical side of the business.
A difficult week for neoclouds
The round was announced as listed neocloud CoreWeave's shares fell to a 12-month low and as Elon Musk predicted heavy consolidation among smaller AI cloud providers, TechCrunch reported. Puklowski called fears of a collapse massively overstated and said customers call him every week asking for more compute on any chip.
SambaNova chief executive Rodrigo Liang told TechCrunch that every company in the market is selling speed, and that the real question is whether it can make money doing so.
Company website: generalcompute.com