Why the first GPU financiers are turning to inference chips in a $400 million deal
General Compute, an AI inference cloud startup, has secured a $400 million loan from investment firm Upper90, in what may be the first financing deal to use inference-specific chips as collateral. The move signals a shift in how AI infrastructure is funded, as investors respond to concerns about the cost of AI tools by backing hardware that runs already-trained open source models more cheaply than cutting-edge chips used to train frontier models from scratch.
General Compute, founded by CEO Finn Puklowski and CTO Jason Goodison, raised a $15 million seed round in May to build its "neocloud" using SN50 chips from Intel-backed SambaNova, which it says offer 16 times faster inference than GPU-based clouds while being cheaper to deploy. Upper90 co-founder Billy Libby, who pioneered GPU-backed lending with data centre firm Crusoe in 2021 before the model became standard practice via companies like CoreWeave, sees inference as the next frontier now that GPU financing is well established. The deal reflects growing investor interest in open source AI infrastructure, alongside rivals such as OpenRouter, Fireworks and TensorWave, and chipmakers including Groq and Cerebras, as alternatives to Nvidia's ecosystem gain traction.
- General Compute secures $400m loan backed by inference chips, not GPUs.
- Upper90, an early GPU financing pioneer, is behind the deal.
- Signals growing investor bet on cheaper, open source AI inference infrastructure.