Sunrise, a Chinese artificial intelligence (AI) chip spinoff from Hong Kong-listed AI software provider SenseTime, has secured a strategic financing of nearly 3 billion yuan ($431.4 million) in a span of 12 months.
A wide range of corporate investors, professional private markets investment firms, and state capital investment platforms participated in its strategic financing. IDG Capital, Soul Capital, Gaorong Ventures, CICC Capital, Puhua Capital, Green Pine Capital Partners, and Yuexiu Industrial Fund were among the investors, just to name a few.
Industrial players like Hong Kong-listed AI company Phancy Group, also known as 4Paradigm; Chinese heavy equipment manufacturer Sany Group’s investment arm HX Capital; green energy supplier GCL Technology; the state-owned Hangzhou Finance and Investment Group; and others also backed Sunrise in the round.
Hangzhou-based Sunrise plans to invest the newly raised capital in the R&D, mass production, and ecosystem building for its next-generation, high-performance GPUs (graphics processing units) specifically optimised for AI inference, the company announced.
It aims for lower power consumption and cost per token, with a goal to achieve greater efficiency AI inference—which is the operational phase of machine learning where a pre-trained AI model applies learned knowledge to generate outputs such as text, images, and answers.
Founded in 2020 as the chip division of SenseTime, Sunrise officially spun out from the parent group in 2024 for independent operation focusing on the R&D and commercialisation of high-performance GPUs and multimodal AI inference chips.
Unlike general-purpose processors that handle both AI training and inference, Sunrise’s GPUs focus exclusively on inference workloads. It believes that this “all-in-AI inference” strategy allows its products to optimise for high-power efficiency, lower costs, and high-memory bandwidth to enable specialised tasks across various industries including smart manufacturing, smart finance, smart healthcare, green energy, and entertainment.
Having invested about 2 billion yuan ($287.6 million) in R&D over the years, Sunrise has mass-produced two AI inference chip products, including S1, a cloud-edge visual inference chip model tailored for image and video analysis; and S2, which is compatible with Nvidia’s CUDA ecosystem, supporting mainstream large models.
The company plans to debut its third chip model S3 in 2026, which it said will offer significant “cost-effectiveness in AI inference.”
“If we can reduce AI inference costs by 90% with more stable services, our competitive edge will be more than ‘just launching another new chip,’ but rewriting the P/L statement of China’s AI industry,” said co-CEO Wang Zhan, a founding member of Baidu who now oversees Sunrise’s commercialisation and fundraising.
“Our goal is to drastically lower the costs and barriers of large model inference, making inference-focused computing power both affordable and widely accessible, thereby fully unleashing the full potential of AGI (artificial general intelligence),” said Wang.
