Source: SGLang project becomes RadixArk worth $400M 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 speed up AI model training, have transitioned to a recently launched commercial startup. The company, it is called RadixArkannounced at the end of August.

RadixArk, which originated from SGLang in 2023 at the UC Berkeley lab of Databricks 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 major contributor to SGLang and former engineer at xAI, is leaving Elon Musk’s AI startup to become co-founder and CEO of RadixArk, according to a LinkedIn announcements he made it last month. Sheng was previously a research scientist at Databricks.

Ying Sheng, Accel, and Lip-Bu Tan of RadixArk did not respond to requests for comment.

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

vLLM, a more mature project to optimize inference, has also moved from an open-source project to a startup. The newly formed company is in talks to raise more than $160 million in funding at a valuation of about $1 billion, Forbes reported last month.

Three people familiar with the deal told TechCrunch that Andreessen Horowitz led the investment into vLLM, though the final amount of the investment remains to be seen. Andreessen Horowitz declined to comment. vLLM founder Simon Mo labeled the information about this round ‘inaccurate’ in a statement to TechCrunch, although he refused to specify any details that were incorrect.

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

A number of large tech companies are already running inference workloads using vLLM, and SGLang has also gained significant popularity over the past six months, Brittany Walker, general partner at CRV, told TechCrunch. His company does not support either company.

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

While most tools remain free, RadixArk has started charging for hosting services, people familiar with the company told TechCrunch.

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