Orient Securities has published a research report stating that 2026 is set to become the first year of mass production for embodied intelligence. The firm observes that model capabilities and scenario-specific know-how among embodied intelligence hardware makers are improving at an accelerating pace, which is expected to expedite commercial breakthroughs across the sector. The brokerage believes that manufacturers actively deploying in various application scenarios, along with full-machine producers boasting stronger model generalization abilities, are positioned to secure greater competitive advantages.
Embodied Intelligence Hardware Makers Are Releasing Rapid Updates, Industry Commercial Breakthroughs Are Expected to Accelerate
Orient Securities notes that 2026 is emerging as the inaugural year of mass production for embodied intelligence. In the first half of 2026, global shipments of humanoid robots surpassed 22,000 units, marking a year-on-year surge of nearly 300%. However, market investors remain deeply divided on the industry's prospects, fearing that current embodied intelligence is still stuck in the "remote-controlled toy" phase, unable to achieve broader market adoption, which could make sustained growth in full-machine shipments unsustainable. Looking at the latest marginal shifts, the research house argues that market sentiment regarding hardware makers' progress in exploring application scenarios is overly pessimistic. The model capabilities and scenario know-how of embodied intelligence hardware producers are improving at an accelerated rate, which should hasten the pace of commercial breakthroughs in the industry.
Rapid Advances in Industry Model Capabilities Are Expected to Expand Potential Application Scenarios
Orient Securities observes that recent improvements in model capabilities within the embodied intelligence industry have been swift, which is likely to broaden the potential application landscape. Previously, Unitree Robotics announced the full open-sourcing of its UnifoLM-WLA-1.0 embodied foundation model. Unitree stated that the model was trained on roughly 2,500 hours of high-quality real-world robot data, enabling a single model to drive robots through 64 desktop manipulation and full-body operation tasks, while also supporting parallel grippers and two types of dexterous hands. Recently, Figure unveiled its new model, Helix 2.5. Figure pre-trained Helix 2.5 on the Index global human behavior dataset and then deployed the model across 30 unseen household environments, reporting that it enabled zero-shot full-body autonomous behavior in robots. Without any new data collection or environment adaptation, the robots' zero-shot success rate soared from 9% to 56%. The research firm believes that Helix 2.5, to a certain extent, validates that embodied intelligence can learn from human experience and transfer those skills to novel environments. Additionally, the industry's growing focus on EGO data is also expected to further accelerate advancements in embodied intelligence model capabilities.
Improving Scenario Know-How Among Embodied Intelligence Makers Is Expected to Drive Shipment Growth
Orient Securities sees that industry applications within embodied intelligence are now flourishing across diverse fronts. The firm attributes this largely to robot manufacturers' continuously deepening grasp of scenario-specific know-how. Embodied intelligence is gradually becoming a contributor to boosting social productivity, which should support sustained growth in shipment volumes. Recently, Qiyuan Robotics, the consumer-grade embodied intelligence brand under Swancor Advanced Materials, hosted a new product launch event, officially selling two personal robots: the Qiyuan Q1 and Qiyuan T1. Both the Qiyuan Q1 and the standard version of the Qiyuan T1 are priced at 19,999 yuan. The Qiyuan Q1 stands 88 centimeters tall and can be folded to fit into a backpack, while the Qiyuan T1 supports switching between wheeled humanoid and quadruped configurations. The brokerage points out that these two home-oriented robots can effectively meet certain users' daily needs in areas such as family education, child companionship, and travel assistance, thereby potentially driving shipment growth. Looking ahead, Orient Securities believes that as embodied intelligence makers enhance their scenario know-how, commercial breakthroughs in the sector are poised to accelerate.
Risk Warnings
Risks include slower-than-expected model development and data accumulation, unclear demand for scenario-based applications, production shortfalls by manufacturers, shifts in national policies, and insufficient industry funding.