Inference cloud operator General Compute raises $400M in debt financing
Artificial intelligence infrastructure startup General Compute Inc. today announced that it secured $400 million in debt financing.
Upper90, the investment firm that is underwriting the round, will initially provide the company with $100 million. General Compute plans to draw down additional funds as customer demand increases.
Unlike most AI infrastructure providers, General Compute is not using Nvidia Corp. graphics cards. The company has based its cloud platform on chips from Advanced Micro Devices Inc. and SambaNova Inc., a fellow startup that recently raised $1 billion in funding.
AI models answer user prompts through a two-step workflow. They start by extracting the meaning of a prompt, a step known as the prefill phase. They then generate a response one token at a time in a process known as decoding.
The decoding phase requires AI models to move a significant amount of data between the host chip’s memory and processing circuits. SambaNova’s accelerators are specifically optimized to speed up that workflow. As a result, they can perform decoding faster than some Nvidia graphics cards.
The speed at which data moves between a chip’s memory and processing modules depends on the physical distance between the circuits. The closer they are to each other, the less time it takes data to complete the trip. The memory and processing circuits in SambaNova’s chips are placed immediately next to one another, which minimizes data travel times.
General Compute uses SambaNova silicon to run decoding calculations. It performs the other step involved in inference workflows, the prefill phase, using AMD’s MI300X graphics card. The chip contains 12 dies and 8 memory stacks that together contain 153 billion transistors.
The MI300X and SambaNova’s chips are designed to run in standard air-cooled server racks. As a result, General Compute can add capacity to its cloud without investing in the pricey liquid cooling equipment that most graphics cards require. That translates into faster infrastructure upgrades.
General Compute enables customers to access its inference hardware through two services.
The first offers managed versions of open-source language models. Developers can access the models through an application programming interface modeled after the one used by OpenAI Group PBC, which eases the learning curve. General Compute’s second service provides dedicated infrastructure environments with more computing capacity.
The company will use its debt financing to expand its cloud. According to a document on its website, General Compute has inked agreements that give it the option to purchase 15 megawatts worth of air-cooled rack capacity in co-location facilities. The company expects that infrastructure to be sufficient for its fourth quarter growth plans.
Photo: Unsplash
Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.
15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more** 11.4k+ theCUBE alumni**— Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network.
About SiliconANGLE Media
theCUBE AIand theCUBE SuperStudios — with flagship locations in Silicon Valley and the New York Stock Exchange — SiliconANGLE Media operates at the intersection of media, technology and AI.
Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.