The two dimensions validated respectively by Unitree and AgiBot, “profitability” on one hand, and “scaled mass production + scenario deployment” on the other, would become the two most important yardsticks for screening the next round of winners. Whether companies can secure both capabilities before the capital‑market window closes will be the central question shaping the competitive landscape in the second half of China’s humanoid‑robot industry.
This report reviews the humanoid robot industry through the first half of 2026, focusing on three key areas: new complete product launches, technological R&D, and strategic deployment. Regarding key components and technologies, it specifically examines the trends and challenges in Battery, VLA model and Tactile Sensing, and provides an in-depth analysis of the strategic developments of several international developers. Finally, the report estimates the Chinese Humanoid Robot Developers’ Total Production in 2026, and provides an industry outlook for 2026, serving as a valuable reference for investment and deployment.
TrendForce combines its expertise in the Foundry/Server industry and leverages it with the supply chain to accurately estimate AI chip shipments, AI Server application ratios, AI major suppliers' analysis, and the adoption of AI memory.
NVIDIA’s integration and deployment for Cosmos, Isaac, and GR00T at GTC 2026 could potentially reduce training cost alongside breakthroughs of simulation technology, as well as realize commercialization and mass production for humanoid robots.
Cloud giants are drastically expanding Capex for AI infrastructure, driving explosive demand for GPU/ASIC racks. However, immense computing demand faces structural component shortages. Meanwhile, OEMs are transitioning into AI system integrators targeting sovereign clouds.
The “Humanoid Robot and Embodied Intelligence Standards System” is not just a technical document, but also a part of China's policy layout, with core purposes including the guidance of a healthy and orderly development of the industry, prevention of industrial fragmentation, and ensuring that safety and ethical standards are prioritized to reduce commercialization risks.
North American CSP CAPEX expansion drives dual demand for NVIDIA and custom chips. ODMs like Foxconn benefit from rack solutions and are expanding US/Mexico production. Power leaders Delta and LiteOn are shifting toward high-voltage, liquid cooling, and system-level integration to support high-density computing.
Tesla CEO Elon Musk plans to complete the construction of mass production lines for Optimus Gen 3 within 2026, targeting an annual output of 1 million units by year-end. However, considerable uncertainty remains regarding Optimus’ high-precision application scenarios, cost structure, supply chain dependency, and production progress—compounded by a history of missed production targets.
Looking ahead to 2026, major global CSPs are significantly increasing capital expenditures on AI infrastructure, focusing on high-density GPU racks and self-developed ASICs to optimize costs. North American giants are leading the adoption of NVIDIA solutions and proprietary chips, driving the share of AI servers and ASICs. Meanwhile, OEMs like Dell and IEIT are aggressively deploying AI solutions amidst intense competition and geopolitical challenges.
Humanoid robot manufacturers 1X Technologies and Boston Dynamics have both enhanced their products' adaptability to unstructured environments through model updates in mid-to-late January 2026. Looking at the current status of the overall industry, the maturity of capabilities among AI models is becoming the primary driving force, as hardware development converge in specifications. However, like AI in other fields, there are also limitations and risks when applied to humanoid robots, which scale with the effectiveness of their training.