Infinity, an AI infrastructure startup, closed a $15 million Series A round at a $100 million valuation. Touring Capital led the investment, with participation from Principal VC and individual backers including researchers from OpenAI and Anthropic.

The funding signals growing investor appetite for inference optimization tools. Infinity builds software that reduces computational costs and latency when running large language models in production. The company addresses a real pain point for enterprises deploying AI applications at scale, where inference expenses often dwarf training costs.

The presence of OpenAI and Anthropic researchers as investors speaks to the startup's credibility within the AI community. These individuals understand the technical challenges of deploying LLMs efficiently and see commercial opportunity in Infinity's approach. Their participation also suggests the company has early product-market validation among sophisticated users who understand inference bottlenecks intimately.

Touring Capital has positioned itself as an infrastructure-focused investor, backing companies solving foundational AI problems. The fund understands that AI adoption will be constrained by cost and performance barriers, making companies like Infinity strategically valuable to the broader ecosystem.

Infinity operates in a competitive space. Companies like vLLM, SGLang, and others have released open-source inference optimization tools. However, Infinity's commercial model and enterprise focus differentiate it. Startups that can package open-source innovations into reliable, managed services for companies unwilling to operate these systems internally have significant runway.

The $100 million valuation appears reasonable for an early-stage infrastructure company with investor backing from domain experts. Series A rounds in AI infrastructure typically land between $50 million and $200 million valuations depending on traction and competitive positioning.

Infinity's path forward hinges on converting AI teams evaluating inference solutions into paying customers. The inference layer has become a defensible moat. Companies that make deploying and running L