Sources: project SGLang spins out as RadixArk with $400M valuation as inference market explodes

Some of the team responsible for maintaining SGLang, a popular open-source tool used by companies like xAI and Cursor to accelerate AI model training, has transitioned to its recently launched commercial startup. That company, called RadixArk, was announced last August.

RadixArk, which originated as SGLang in 2023 inside the UC Berkeley lab of Databricks co-founder Ion Stoica, was recently valued at about $400 million in a round led by Accel, according to two people familiar with the matter. TechCrunch could not confirm the size of the funding.

The startup previously raised angel capital from investors, including Intel CEO Lip-Bu Tan, the people said.

Ying Sheng, a key contributor to SGLang and a former engineer at xAI, left Elon Musk’s AI startup to become the co-founder and CEO of RadixArk, according to a LinkedIn announcement she made last month. Sheng was previously a research scientist at Databricks.

RadixArk’s Ying Sheng, Accel, and Lip-Bu Tan did not respond to a request for comment.

Both SGLang and RadixArk focus on optimizing inference processing — essentially allowing models to run faster and more efficiently on the same hardware. Together with model training, inference represents a large portion of the server costs associated with AI services. As a result, tools that optimize the process can create enormous savings almost immediately.

vLLM, a more mature project for optimizing inference, has also been transitioned from an open-source project to a startup. The newly formed company has had conversations about raising upwards of $160 million in funding at a valuation of about $1 billion, Forbes reported last month.

Three people familiar with that deal tell TechCrunch that Andreessen Horowitz is leading the investment into vLLM, though the final numbers of that investment remain to be seen. Andreessen Horowitz declined to comment. vLLM co-founder Simon Mo characterized the information about this round ‘factually inaccurate’ in a statement to TechCrunch, though he declined to specify which details were incorrect.

Like SGLang, vLLM was incubated in Ion Stoica’s lab at UC Berkeley. Stoica, a professor at UC Berkeley, is the famed co-founder of Databricks as well as a number of other startups.

Several large tech companies already run their inference workloads using vLLM, and SGLang has also gained significant popularity over the last six months, Brittany Walker, a general partner at CRV, told TechCrunch. Her firm did not back either company.

RadixArk is continuing to develop SGLang as an open-source AI model engine. The startup is also building Miles, a specialized framework designed for reinforcement learning, which allows businesses to train AI models to become smarter over time.

While most of its tools remain free, RadixArk has started charging fees for hosting services, a person familiar with the company told TechCrunch.

Startups providing inference infrastructure for developers have seen a surge in funding in recent months, underscoring the continued importance of the inference layer for AI. Baseten recently secured $300 million at a $5 billion valuation, the Wall Street Journal reported on Tuesday.  This follows a similar move by rival Fireworks AI, which raised $250 million at a $4 billion valuation last October.