Daxiao Robotics Raises Hundreds of Millions in H1 2026, Launches Kairos-HomeWorld Whole-Home AI Training Framework
SenseTime-affiliated Daxiao Robotics has accumulated hundreds of millions in funding during H1 2026, backed by Ant Group, Geely Capital, and Lenovo Ventures. The company's latest breakthrough, Kairos-HomeWorld, is the first AI framework to generate structurally complete, physically plausible whole-home 3D environments — solving a critical data bottleneck for training domestic service robots on Chinese home layouts.

SenseTime-affiliated Daxiao Robotics has raised hundreds of millions of RMB in H1 2026 and launched Kairos-HomeWorld — the first AI framework to generate physically plausible, structurally complete 3D home environments for training domestic service robots.
Daxiao Robotics has emerged as one of the most consequential bets in China's domestic service robotics race. Backed by Ant Group, Geely Capital, and Lenovo Ventures, the company has accumulated hundreds of millions in funding during the first half of 2026; a financing trajectory that reflects both the strategic importance of its parent company SenseTime and the growing conviction among institutional investors that whole-home robotic automation is the next major frontier in embodied AI.
Kairos-HomeWorld: Solving the Data Bottleneck for Domestic Robots
The centrepiece of Daxiao's June 2026 announcement is Kairos-HomeWorld, a generative AI framework designed to produce structurally complete, physically plausible 3D home environments at scale. One of the most persistent obstacles to training domestic service robots has been the scarcity of high-quality, diverse training data that accurately reflects the layout, object placement, and spatial constraints of real Chinese homes.
Existing simulation environments; including those used by international robotics labs, are predominantly modelled on Western home layouts, which differ substantially from the compact, multi-functional spaces typical of Chinese urban apartments. Kairos-HomeWorld addresses this directly by generating environments calibrated to Chinese residential norms, providing Daxiao's robots with training data that is both physically accurate and culturally relevant.
The SenseTime Advantage and What It Means for Competitors
Daxiao's relationship with SenseTime provides it with a structural advantage that pure-play robotics startups cannot easily replicate. SenseTime's deep expertise in computer vision, spatial AI, and large-scale model training is directly applicable to the challenges of domestic robot perception and manipulation. The Kairos-HomeWorld framework is, in effect, an extension of SenseTime's foundational AI capabilities applied to the physical world.
For competitors in the domestic service robot space, including Unitree's home-focused variants, Leju Robotics, and a growing number of SoftBank-backed entrants, Daxiao's combination of institutional backing, proprietary training infrastructure, and SenseTime's model library represents a formidable moat.
Daxiao Robotics is not building a robot, it is building the data infrastructure that makes domestic robots commercially viable in China. If Kairos-HomeWorld delivers on its promise, the company that controls the training environment may ultimately control the market. The deeper question is whether a world-model advantage translates to hardware dominance, or whether the two remain separable in China's increasingly fragmented robotics supply chain.
Primary Sources:
1. 36Kr / Pencil News (铅笔道) — Shihang Intelligent 1 Billion RMB Series A (June 15, 2026)
2. FutureX · Physical AI Daily, Issue 29 — China Robotics Intelligence Digest (June 16, 2026)
3. Caixin Global — China's Humanoid Robot Boom Sparks Price War Fears (June 15,2026)
4. Gasgoo ACE Robotics Intelligence — China Robotics Industry Coverage (June 15, 2026)
5. SenseTime Group — Corporate AI Capabilities Overview (June, 2026)
Disclaimer: All editorial content is independently written by RobotAIGeek based on publicly available Chinese and English-language sources. Quoted material is attributed to its original publisher not independently verified by RobotAIGeek.












