Gasgoo Munich- On June 16, SiliconFlow, a third-party Model-as-a-Service (MaaS) platform, announced it has closed a Series B funding round exceeding 2 billion yuan.
The round drew capital from industrial players, financial institutions, and state-backed entities.

Image source: NIO Capital
With this latest round, SiliconFlow has secured strategic backing from leading players across the AI supply chain — spanning energy, chips, computing infrastructure, cloud services, large models, and application scenarios.
Since its inception, SiliconFlow has championed a "Token Factory" model, aiming to democratize AI technology by boosting the production efficiency and large-scale supply of tokens.
Official figures reveal explosive growth in its enterprise business over the past year. Daily token calls have hit the trillions, serving more than 10 million users and 10,000 corporate clients. Revenue surged over tenfold year-on-year, while monthly overseas revenue climbed into the millions of dollars.
Its flagship client roster spans top state-owned enterprises in energy, finance, and transportation, along with major telecom operators and regional intelligent computing centers. It also serves cloud giants, leading large-model firms, and hundreds of AI application developers.
Technically, SiliconFlow has built a fully self-developed token production line. Its proprietary inference engine integrates advanced technologies like PD separation, KV cache management, and expert parallelism. This allows it to stably support mainstream models such as DeepSeek and Qwen across a diverse range of chips, including Nvidia, Ascend, MetaX, and Moore Threads.
Meanwhile, its heterogeneous computing management and scheduling system enables unified intelligent scheduling and elastic scaling across regions. It converts raw computing power into standardized token factories, addressing industry pain points like complex hardware adaptation and low utilization rates.
To date, the platform has rapidly adapted to more than 160 models, covering all modalities.








