General Compute Secures $400 Million Debt Facility to Scale AI Inference Cloud

San Francisco-based General Compute has secured a committed debt facility of up to $400 million from Upper90 Capital Management. The financing begins at $100 million and can increase with customer demand, allowing the company to expand its inference cloud infrastructure built on specialized chips from SambaNova.

The company stated that its infrastructure can deliver AI inference up to sixteen times faster than standard GPU-based systems, with energy use six times lower. Its air-cooled racks require no specialized liquid cooling, enabling deployment within weeks. General Compute reported that it already has over $300 million of secured, price-protected chip supply and expects to be the first to deploy ASIC silicon at scale for inference workloads.

The SambaNova SN40 and SN50 chips used in General Compute’s infrastructure are designed for high throughput token generation while operating on about twenty kilowatts per rack. According to the company, users can switch to its service in less than thirty seconds and run models from providers including OpenAI, DeepSeek, and MiniMax.

Upper90 is also an equity investor in General Compute. The firm stated that its investment aligns both its debt and equity positions to support the company’s growth and the rapid scaling of its AI compute infrastructure.