AI infrastructure company Infinity closed a $15M round at a $100M valuation, with Touring Capital, Principal VC, and individual researchers from OpenAI and Anthropic participating.
The company’s inference engine lets businesses run open-weight models on standard server hardware rather than expensive GPU clusters. Infinity claims its technology reduces large language model memory requirements during inference by up to 80%, making AI serving practical on commodity infrastructure.
Researchers from both OpenAI and Anthropic participated as angel investors in their personal capacity, signaling researcher-level interest in making model inference more accessible as open-weight models gain traction in the enterprise.
Infinity enters a competitive market where larger players like Baseten and Fireworks AI have raised rounds of $1.5B each at valuations above $13B. Infinity differentiates with a purely software-based approach focused on optimization rather than building full-stack infrastructure.
The company was founded by former Google Brain and Meta AI researchers who saw an opportunity between hyperscaler cloud offerings and the need for lightweight inference that can run anywhere. Infinity’s software works on premises, at the edge, and in cloud environments.
The startup plans to use the funding to expand its engineering team, open-source parts of its inference stack, and build a commercial sales operation targeting mid-market enterprises.
