vLLM creators launch Inferact with $150 million seed round

Inferact, a startup commercializing the open-source vLLM project, has raised $150 million in seed funding at an $800 million valuation. The funding round was co-led by venture capital firms Andreessen Horowitz and Lightspeed Venture Partners. In the technology sector, seed funding represents the early-stage capital investment required to transition an open-source project into a commercial enterprise. Inferact is utilizing this capital to focus on inference, which is defined as the process of deploying artificial intelligence models in applications. By optimizing this process, the company aims to make the execution of artificial intelligence workloads faster and more affordable for enterprises. The transition of vLLM from an open-source tool into a venture-backed startup represents a significant step in the commercialization of artificial intelligence infrastructure.

The debut of Inferact closely mirrors the recent commercialization of SGLang, another open-source project designed to improve the efficiency of artificial intelligence systems. SGLang has been commercialized as a startup named RadixArk. The rapid transition of these academic projects into venture-backed companies underscores the intense investor interest in software that optimizes the deployment phase of artificial intelligence. The financial details of these two newly formed entities demonstrate the significant market valuations being established for inference-focused startups:

• Inferact: The company raised $150 million in seed funding at an $800 million valuation to commercialize the vLLM project, with Andreessen Horowitz and Lightspeed Venture Partners co-leading the investment.
• RadixArk: The startup commercializing SGLang secured capital at a $400 million valuation, in an investment round led by Accel that was reported on Wednesday.

Both vLLM and SGLang share a common academic lineage, having been incubated in 2023 at the University of California, Berkeley research lab of Ion Stoica, who is also a co-founder of Databricks. While SGLang is being commercialized through RadixArk, vLLM has already achieved notable adoption among major corporate users. According to Simon Mo, the CEO of Inferact and one of the original creators of the vLLM project, existing users of vLLM include Amazon’s cloud service and the shopping app. This existing user base provides Inferact with an established foundation as it transitions from an open-source project into a commercial entity.

Why it matters

As the artificial intelligence industry shifts its focus from training models to deploying them in applications, technologies that optimize inference are becoming critical infrastructure. Tools like vLLM and SGLang, which make these systems run faster and more affordably, are attracting substantial investor attention as businesses seek to manage operational costs.