General Compute lands $100M Series A to build an inference cloud on Nvidia rivals' chips

What's the deal? General Compute has raised $100 million in a Series A co-led by BlackbirdDealroom has a profile for this one. Try Dealroom → and Abstract, with Fuse Venture PartnersDealroom has a profile for this one. Try Dealroom → and Village Global also taking part. The San Francisco startup builds an inference cloud using specialised chips from Nvidia competitors Cerebras, SambaNova, and Positron AI. The round values the company at $800 million post-money, up from $60 million after a May seed round.

What's the endgame? General Compute finances, deploys, and operates GPUs and purpose-built inference chips, matching workloads to the hardware that runs them best. Customers get dedicated bare metal or a guaranteed tokens-per-second service-level agreement without managing infrastructure themselves. First major deployments go live in the first quarter of 2027, with more than 50MW in the pipeline.

Why now? The raise lands as the neocloud sector faces pressure: CoreWeave's stock sits near a 12-month low, and Elon Musk has predicted "200+ neoclouds are going to see guppy-level death and consolidation." A deal with a large customer worth $100 million in annual recurring revenue paved the way for the round.

How it's built: Rather than constructing data centres from scratch, General Compute leases space, taps smaller power sources, and converts former bitcoin-mining sites. It struck a $400 million deal to buy chips from SambaNova, financed with a debt facility from Upper90Dealroom has a profile for this one. Try Dealroom →. Chief executive officer Finn Puklowski expects to deploy 75MW in 2027.

What could go wrong? The model still demands heavy capital. "Capital is going to be a full-time job for me, both on debt and equity," Puklowski told TechCrunch. The 13x valuation jump and reported customer contract remain unproven until deployments and revenue arrive.

The signal: At $100 million, this ranks among the largest 1% of all-time US enterprise software Series A rounds. It is a bet that specialised inference silicon — not GPU-stuffed hyperscale build-outs — is where AI compute demand is heading.

Read more: generalcompute.com, TechCrunch

Image credit: General Compute