DeepCtrls, a company engaged in physical artificial intelligence, has completed a Series B funding round, raising several hundred million yuan. These funds are intended to scale its physical AI platform for the needs of global energy infrastructure, serving over 360 corporate clients.
Funding and Investors
Jinko Energy acted as a strategic investor in this round, with SDIC Innovation and CMB International as co-leads. Existing investors, including Sequoia China, Source Code Capital, Guangyuan Capital, and CMC Capital, were also among the participants.
Technological Advantage
DeepCtrls develops hybrid physics and AI models that combine an understanding of physical mechanisms with deep learning. This allows it to solve the problem of distrust in 'black boxes,' which has long hindered the adoption of industrial AI. Unlike most industrial AI systems that operate only at the diagnostic or consulting level, DeepCtrls claims its AI is capable of achieving autonomous closed-loop control at L4/L5 levels, surpassing standard market systems based on L2/L3 rules.
Company History and Product
The company was founded in 2018 by Dr. Li Hui. He spent over two decades linking physics modeling at Tsinghua University with AI and energy research at the Lawrence Berkeley National Laboratory. Thanks to this, DeepCtrls created the PhyAI engine, which was trained on real-time operational data from over 300,000 devices. This engine demonstrates a generalization error below 3%, approaching the theoretical limit of system optimization.
Clients and Development Plans
Over 360 enterprises have utilized the company's products, including TSMC, CXMT, Zhongji Innolight, Foxconn Industrial Internet, Tencent, ByteDance, and CATL. The DeepBot product, a standardized physical AI controller, is focused on optimizing energy consumption in industrial manufacturing, data centers, and commercial buildings. The company's financial figures show a year-on-year doubling of revenue since commercialization began in 2023, along with achieving scale profitability last year, which is a rare achievement for industrial AI.
International Expansion
The raised investments will be directed towards further research and development, expanding the DeepBot product, and entering international markets, including Southeast Asia, the Middle East, Europe, and North America. In 2024, DeepCtrls began global expansion, deploying its products in more than ten countries and regions. Chief Operating Officer Zhang Yuxin noted that overseas data centers facing thermal and energy management challenges represent a particularly promising area for growth in physical AI, as most foreign energy systems still operate at the L2 expert rule control level.
