SiliconFlow Raises $330M for AI Inference Engine Expansion

SiliconFlow, a prominent Chinese AI infrastructure provider, has successfully closed a substantial Series B funding round, raising over $330 million USD (approximately 2 billion RMB). The significant capital infusion was led by a consortium of strategic investors including JinkoSolar, NIO Capital, and Jiyuan Capital, alongside over a dozen other entities from the energy, semiconductor, cloud services, and finance sectors. This funding is earmarked for critical advancements in their inference engine technology, enhancing compatibility with domestic Chinese chip architectures, and accelerating international market expansion.

The company positions itself as a vital "Token Factory," specializing in packaging computing power, algorithms, and system capabilities into standardized, scalable inference services. SiliconFlow operates as a third-party Model-as-a-Service (MaaS) platform, bridging the gap between large model developers and the essential computational resources they require. Their offering aims to democratize access to low-cost, high-performance inference for enterprises and developers navigating the complex AI ecosystem.

SiliconFlow's proprietary inference engine already boasts support for more than 160 diverse AI models. This includes prominent large language models (LLMs) such as Alibaba Group's 'Qwen', DeepSeek's foundational models, Zhipu AI's flagship 'GLM', and Moonshot AI's 'Kimi'. The company has demonstrated a commitment to hardware integration, notably launching DeepSeek inference services on Huawei's 'Ascend' chips in February 2025 and releasing their 'Elastic GPU' computational resource scheduling engine in April 2026.

The AI infrastructure market, particularly the MaaS segment, is experiencing explosive growth. Global spending on AI infrastructure is projected to reach hundreds of billions of dollars in the coming years, driven by the insatiable demand for training and deploying increasingly sophisticated AI models. SiliconFlow's focus on optimizing inference, a critical bottleneck for real-world AI applications, places it at the forefront of this expanding market. According to IDC, SiliconFlow is already recognized among the top four players in China's public cloud MaaS market, underscoring its strong domestic traction.

Recent performance metrics highlight SiliconFlow's rapid ascent. The company reported a daily average of trillions of token calls within the enterprise market over the past year. Their service now reaches over 10 million individual users and more than 10,000 enterprise clients. Revenue has surged by more than tenfold year-over-year, with international markets contributing several million dollars in monthly sales, indicating significant global potential. This rapid scaling reflects the broader trend of Chinese AI models gaining substantial API usage, even surpassing US counterparts in certain metrics.

This funding round positions SiliconFlow to further solidify its competitive edge. By investing in inference engine upgrades and ensuring seamless integration with China's burgeoning domestic semiconductor industry, the company is strategically aligning itself with national technological priorities. Simultaneously, its push into overseas markets signals ambitions to become a global leader in AI inference solutions, capitalizing on the worldwide demand for efficient and accessible AI deployment.