RadixArk Raises Accel-led $100M Seed Round

• RadixArk raised $100M at a $400M valuation, led by Accel and Spark Capital
• Founded by Ying Sheng and Banghua Zhu, it builds AI infrastructure powered by SGLang
• The platform handles trillions of tokens daily and supports end-to-end AI development

RadixArk has secured $100 million in initial investments, valuing the company at $400 million. Investments came primarily from Accel and Spark Capital with participation from NVentures and other investors.

Founded by two experts in the space of AI infrastructure and modeling, Ying Sheng and Banghua Zhu, both previously worked at xAI and Nvidia. Their prior experience building large-scale AI systems has influenced the vision of RadixArk to make AI infrastructure more accessible with a focus on optimized performance.

RadixArk is an AI infrastructure startup focused on creating open and scalable systems for the training, deployment, and execution of advanced AI models. The core of the platform is the SGLang, an open-source inference engine built to enable optimizing the way AI models interface with hardware through the use of the SGLang as a middle layer between the model and the compute system, which reduces memory footprint and improves the efficiency of the interactions, allowing for faster and lower-cost AI implementations.

This technology is already being utilized at a global level, processing trillions of tokens on an ongoing basis for both enterprise and research applications.

By providing a comprehensive platform for developing AI, RadixArk is excited about going beyond traditional inference platforms to support the entire lifecycle of an AI project, including model creation (training), fine-tuning/optimizing, and reinforcement learning, along with large-scale deployment.

With this initiative, RadixArk wants to make advanced AI infrastructure more easily available so that developers, entrepreneurs, and larger organizations are better able to develop and operate advanced AI systems with less effort.