SGLang Evolves into RadixArk with $400 Million Valuation

SGLang, an open-source AI tool originally developed as a research project at the University of California, Berkeley, has transitioned into a standalone company named RadixArk. This new venture has secured a valuation of approximately $400 million. The transition marks a significant milestone for the project, which was spearheaded by Ion Stoica, a co-founder of Databricks, in 2023.

RadixArk aims to enhance the inference stage of AI models. This crucial phase is where trained models generate answers or results, leading to reduced computing costs and improved performance. The rise of RadixArk coincides with increasing demand for tools that optimize AI efficiency.

Funding and Growth

According to reliable sources, the startup raised funds from notable investors, including the venture capital firm Accel and early angel investors like Intel CEO Lip-Bu Tan. RadixArk was publicly announced in August 2023, and the team has since fully transitioned away from its origins as part of an academic project.

RadixArk’s focus is on making large AI models operate faster and more cost-effectively. Organizations such as xAI and Cursor have utilized SGLang to accelerate both training and inference processes within their AI systems. The departure of part of the original SGLang team from the UC Berkeley lab to join RadixArk reflects the growing interest in performance-enhancing AI tools.

Future Developments

The company plans to continue developing SGLang as an open-source engine while also introducing new tools. One such initiative is Miles, a reinforcement learning framework designed to facilitate the continuous improvement of AI models over time. While RadixArk intends to keep much of its software free, it has started offering paid hosting services.

This shift signifies a broader trend in the AI infrastructure landscape, where tools initially conceived as academic or open-source projects are evolving into commercial entities due to heightened demand. Companies concentrating on AI inference, which involves enabling models to operate in real time, are increasingly attracting investment as the market expands.

As RadixArk continues to develop its offerings, it represents a significant evolution in the intersection of academia and commercial enterprise, showcasing how innovative ideas can successfully transition into thriving businesses.