344
Productivity & Workflow355
Automation & Workflow224
Software Development251
Marketing & Growth192
AI Infrastructure & MLOps174
Writing & Content Creation203
Data & Analytics141
Design & Creative170
Photography & Imaging156
Customer Support131
Sales & Outreach125
Voice & Speech135
Education & Learning131
Operations & Admin87
Upper90 has given General Compute a $400 million loan, using AI inference chips as collateral. It signals growing interest in cheaper ways to run AI.
In short: General Compute secured a $400 million loan from Upper90, and the loan is backed by chips designed to run AI models.
General Compute, a startup that sells cloud computing for AI, has landed a $400 million loan from investment firm Upper90. The loan is backed by the company’s AI inference chips, meaning the chips act like collateral (similar to how a house backs a mortgage).
“Inference” is the part of AI where a trained model is used to answer questions or generate text, images, or code. It is different from “training,” which is the expensive process of building the model in the first place. In simple terms, training is like writing and editing a cookbook, and inference is like using the cookbook to cook meals.
General Compute is building its service around chips from SambaNova. The company says its SN50 chips are built for inference, use less power, and do not need water cooling systems. General Compute also claims these chips can deliver much faster inference than many GPU-based (graphics chip) clouds.
Upper90’s CEO, Billy Libby, said his firm previously helped finance large GPU purchases starting in 2021. The TechCrunch report notes that chip-backed loans have become more common as lenders got more comfortable valuing advanced chips.
Many AI services are expensive to run, and those costs can show up in higher prices for businesses and consumers. A shift toward inference-focused chips could make everyday AI features cheaper to operate, especially for services using open-source AI models (models that anyone can use and build on).
Source: TechCrunch AI