Inferact launched Wednesday with $150 million in seed funding at an $800 million valuation to commercialize vLLM, the open-source inference engine that reduces AI deployment costs by up to 70%. The round, co-led by Andreessen Horowitz and Lightspeed Venture Partners, represents one of the largest seed valuations ever and signals a fundamental shift in AI industry priorities from model training to deployment optimization.

The infrastructure bottleneck is real. Organizations deploying AI applications are discovering that inference costs—the expense of running trained models to generate outputs—often exceed training expenses over a product’s lifetime. Companies like Stripe report 70% cost reductions using vLLM, while the technology enables significantly faster processing and higher hardware utilization across the AI stack.

Scrunch AI secured $15 million in Series A funding to build infrastructure that makes the internet readable by AI agents, addressing a fundamental bottleneck where most enterprise websites remain inaccessible to autonomous systems.

The Salt Lake City-based company emerged from a market reality check: while AI agents proliferate across enterprise workflows, the underlying web infrastructure was designed for human browsers, not machine consumption. This creates a discovery and interaction barrier that Scrunch aims to eliminate through its Agent Experience Platform (AXP).