GPU Startup Andromeda Reaches $1.5B Valuation After Paradigm Investment

Quick Facts

• Paradigm invested $60 million total in Andromeda, bringing the GPU rental startup to a $1.5 billion valuation
• Andromeda manages up to 100,000 GPUs through its platform that consolidates multiple providers into one interface
• The company originated as a side project by investors Nat Friedman and Daniel Gross before spinning out as an independent startup

Andromeda AI Inc. closed a funding round at a $1.5 billion valuation, with Paradigm providing $60 million in total investment to date. The startup helps companies rent artificial intelligence infrastructure through a unified platform.

The company operates with about 20 employees and recently opened a San Francisco office. Andromeda expects to manage up to 100,000 GPUs this year across its network of hardware providers.

“I tell the team we need to be able to provide the reliability and scale of a hyperscaler, with the speed and flexibility of the emerging market,” said CEO Wil Moushey. “Our competition is unutilized GPUs.”

Andromeda vets GPU providers before adding them to its platform, checking performance and security requirements for both graphics cards and supporting equipment like storage arrays. Customers can purchase capacity from multiple providers through one interface with consolidated billing.

The startup began as a side project by investors Nat Friedman and Daniel Gross at their firm NFDG. They committed over $100 million upfront and accumulated more than 4,000 GPUs by early 2024. Portfolio companies like voice startup ElevenLabs and video generation company Pika used the infrastructure.

Paradigm co-founder Matt Huang sees potential for GPU compute to become a standardized commodity similar to wheat trading. “This is a massive new market that really needs a lot of infrastructure to help it work smoothly,” Huang said.

One anonymous AI startup founder who uses Andromeda said the company “basically serves as this excellent engineering team for us,” handling GPU cluster debugging while allowing clients to focus resources elsewhere.

Andromeda could support financial derivatives for GPU capacity, according to Moushey. “You can’t run a $50 million training workload on a financial derivative,” he said, positioning the platform as underlying infrastructure for potential GPU futures markets.

The funding reflects explosive growth in GPU demand, with infrastructure representing 40-60% of typical AI project budgets. Morgan Stanley estimates 70% of GPU cloud spending will shift to inference workloads by late 2026.

Read more: On-demand GPU startup Andromeda raises funding at $1.5B valuation