Daimon Robotics raises hundreds of millions in Ant Group-led round

Ant Group led a strategic financing round of several hundred million yuan in Daimon Robotics, whose tactile world model claims 2x the task success rate of rival π0.5 in contact-intensive operations.

The company said the round, announced Aug. 11, follows a 100-million-yuan Series A closed two months ago, with existing shareholders oversubscribing. Daimon has now assembled a roster of industrial investors including China Merchants Capital, Lenovo Capital and Incubator Group, Inovance Industrial Investment, China Mobile and China Telecom.

Daimon's Daimon-TWM model, launched recently, uses native tactile sensing across understanding, reasoning, prediction and verification. At 10B parameters, it runs real-time inference on NVIDIA RTX 5090 hardware and operates a 100Hz high-frequency control loop. The company claims the model achieves a 2x average success rate over π0.5 in undisturbed conditions and up to 10x improvement under disturbance.

The funding extends Ant Group's reach into embodied intelligence, where tactile data is emerging as the next competitive frontier. Daimon has served more than 200 customers globally, including over 50 overseas, with deliveries to OpenAI, Figure, Physical Intelligence, Skild AI, Meta, BMW and Google DeepMind. The company is also entering supply chains of industrial automation firms including Inovance and Tesla suppliers.

From CMU to Shenzhen: Building a Tactile Stack

Daimon's technical lineage traces to Carnegie Mellon University, where Professor Matthew T. Mason's robotics operation research laid the theoretical foundation for dexterous manipulation. Wang Yu, Mason's first PhD graduate, later co-founded the Robotics Institute at Hong Kong University of Science and Technology and brought the academic system to China. At the end of 2023, Daimon launched operations in Shenzhen with Wang Yu as co-founder and chief scientist, alongside three PhD graduates he trained: CEO Duan Jianghua, Chief AI Scientist Yuan Weihao, and R&D head Du Yipai.

The company's bet is that tactile perception — not vision alone — will drive the next phase of embodied intelligence. Industry research, including work by Stanford's Li Fei-Fei team, has shown that tactile data can significantly improve robot task success rates and data utilization efficiency. Daimon's Daimon-TWM architecture embeds tactile tokens natively into the model rather than splicing them as additional inputs.

The Data Flywheel

Daimon has built Daimon-Infinity, which it calls the world's largest multimodal physical world dataset containing tactile data, with hundreds of thousands of hours of operation data and plans to expand to millions of hours within the year. The first 10,000 hours were open-sourced and received nearly 5 million downloads on Alibaba's ModelScope community, ranking first among all embodied datasets.

The company is also scaling data collection through a partnership with China Mobile, deploying the world's first "Embodied Data Collection 5S Store" in Chenzhou, Hunan. The first phase plans 1,000 sets of equipment, expected to generate 1 million hours of real operation data per year at full capacity.

Daimon claims eight "global No.1" metrics, including the first to achieve 10,000-piece-level shipment of vision-tactile perception equipment, the world's No.1 shipment volume in the category, and the world's No.1 revenue scale in the vision-tactile track.

For investors, the question is whether tactile intelligence can sustain the scaling curve that pure vision approaches have shown. Daimon's model, at 10B parameters, is designed for cross-body and cross-configuration deployment, potentially reducing the cost of adapting robots to new tasks. The company's customer list — spanning OpenAI, Meta, BMW and Google DeepMind — suggests demand is broadening beyond research labs into industrial applications.

This article is for informational purposes only and does not constitute investment advice.