AI inference cloud startup General Compute has received a $400 million loan from technology investment firm Upper90. This may be the first transaction to offer inference-specific chips as collateral. These are chips built to run already trained AI models quickly and efficiently, rather than the more expensive chips used to build the models in the first place.
The funding is the latest sign that the market is responding to concerns about the price of AI tools and tokens by turning to infrastructure that runs open-source models more cheaply than Frontier Institute’s latest LLM.
General Compute, founded by CEO Finn Puklowski, raised a $15 million seed round in May from Intel-backed chipmaker SambaNova to build an inference neocloud around silicon. (Neocloud is purpose-built for AI workloads, unlike the general-purpose infrastructure provided by traditional hyperscalers like AWS or Azure.)
The company’s SN50 chip is designed for inference. It is power efficient and does not require expensive water cooling systems. This means they can be deployed faster than GPUs across a variety of data centers. General Compute says the new chip will deliver inference 16 times faster than GPU-based cloud.
The challenge is getting a lot of these chips, especially if you’re a new company.
Upper90 co-founder and CEO Billy Libby, a former Goldman Sachs quant trader, had a playbook for this. In 2021, his company funded the purchase of GPUs for energy-focused data center startup Crusoe, which he believes is the first loan of its kind to the value of advanced chips.
At the time, traditional lenders avoided these transactions due to the risk and uncertainty of GPUs declining in value. But after CoreWeave made chip-backed lending its business model and the basis for its blockbuster IPO, this type of financing became common.
“When we funded the first group of Nvidia GPUs, the market was inefficient,” Libby told TechCrunch. “We were able to put something together as early participants and reap the rewards of the risk.”
Now that GPUs are relatively well understood and perhaps overbought, Upper90 is turning to companies like General Compute to ride the next wave of the AI boom. “We thought an open source model would be important and we looked for players to engage in inference last year,” Libby said. “Not everyone needs a supercomputer, but they do need inference and AI.”
This argument is growing stronger as companies providing access to open models like OpenRouter and Fireworks attract new investment at enormous valuations. New models, such as Kimi’s K3, which recently launched this week, have proven to compete with recent releases from Anthropic and OpenAI in coding benchmarks. And new chipmakers like Groq and Cerebras have attracted interest from both underwriters and the public market.
General Compute’s ability to access chips outside the Nvidia ecosystem is also important for the same reason. TensorWave, another AI infrastructure company, is making a similar investment in its partnership with AMD. As more alternatives to Nvidia emerge, compute providers not tied to the Nvidia deal may have an advantage in providing cost-effective inference.
“Chips that have amazing (total cost of ownership) or can run much faster than Nvidia are starting to expand, but there aren’t a lot of buyers,” Puklowski said. “Being with the Upper90 doesn’t just mean ‘a cool startup got money to buy computers.’ Likewise, it’s the first sign that capital is organizing itself and Nvidia’s monopoly dominance is fragmenting.”
If you purchase through links in our articles, we may receive a small commission. This does not affect our editorial independence.
