RadixArk, the company behind the open-source tool SGLang, has recently achieved a valuation of $400 million following a funding round led by Accel. This valuation is particularly impressive for a startup that launched just last August. RadixArk emerged from UC Berkeley's lab, founded by Ion Stoica, and has attracted significant talent, including Ying Sheng, a former engineer at Elon Musk’s xAI, who is now the co-founder and CEO. SGLang is gaining traction among major companies like xAI and Cursor, as it optimizes AI model training and inference processing, which are key drivers of operational costs in AI services. The trend of open-source tools transitioning into venture-backed startups is becoming more common. Another example is vLLM, a project incubated in the same UC Berkeley lab, which is reportedly in talks for $160 million in funding at a valuation of around $1 billion. While vLLM has established itself as a more mature player in the inference optimization space, RadixArk is still developing SGLang as an open-source engine and is also working on a new framework called Miles for reinforcement learning. Both companies highlight the growing importance of inference infrastructure in AI, a sector that has seen a surge in funding recently. For example, Baseten raised $300 million at a $5 billion valuation, while Fireworks AI secured $250 million at a $4 billion valuation last October. RadixArk has begun charging for hosting services, indicating a shift towards monetization while continuing to support its open-source roots.