Andromeda

Andromeda AI Inc. operates a marketplace and brokerage for GPU compute, matching AI teams that need training and inference capacity with data centers, telecom and internet providers, cryptocurrency miners and cloud operators that hold unused GPUs. Buyers define a workload (GPU type, quantity, region, timeline); the platform sources, benchmarks and prices capacity across more than 100 providers in real time, then delivers it under standardized SLAs and contracts with a single invoice, one support channel and observability tooling. Suppliers onboard their infrastructure, pass an automated certification engine that validates performance against enterprise-grade benchmarks, and receive routed demand executed on standardized terms. [0][1][6]

The company positions itself as market infrastructure rather than a compute provider, describing an aspiration to build a liquidity layer for global AI compute and to enable commodity-style trading of GPU capacity analogous to oil or wheat futures. Before listing a provider, Andromeda verifies that GPUs and supporting equipment such as storage arrays meet performance and security requirements, and its engineers monitor infrastructure for reliability issues during training runs. It also publishes the Andromeda Pricing Index, which reports current going rates for different GPU types including region-specific variation, which customers can use in price negotiations. [5][6][7]

Andromeda began as the Andromeda Cluster, a managed GPU cluster set up by investors Nat Friedman and Daniel Gross to give early-stage startups access to scaled AI infrastructure, and has since spun out as an independent startup. It works with AI labs, data centers and cloud providers, routing training and inference jobs across global supply. [5][7]