Andromeda Cluster operates a global GPU compute marketplace that connects AI teams with high-performance infrastructure across dozens of providers worldwide. Founded in 2022 by Nat Friedman and Daniel Gross, the company handles the sourcing, benchmarking, pricing, certification and deployment of GPU compute, so that researchers and engineers can focus on model development rather than negotiating hardware contracts. More than a billion GPU-hours have flowed through the network to date.
The firm began as a single multi-thousand-GPU cluster built to support the founders' venture portfolio. Demand quickly outgrew that initial offering, and the operation expanded into a marketplace spanning dozens of infrastructure providers globally. Its stated aim is to become what it calls the liquidity layer for global AI compute, comparable to the electrical grid for manufacturing, with an emphasis on the benchmarks, standards and trust mechanisms that make compute transactions efficient.
Technical work spans GPU compute infrastructure, AI and ML training systems, hardware benchmarking, resource provisioning and deployment, and the market and pricing systems underpinning the marketplace. Engineering effort is directed at infrastructure problems, at a scale ranging from individual clusters to a worldwide provider network serving AI startups, researchers and frontier model developers.
The company frames its mission around removing hardware-access bottlenecks that it argues have kept startups and researchers from frontier-scale AI development. Its work is oriented toward early-stage teams and builders rather than hyperscalers, and its culture is described as engineering-led and long-term in ambition.
