On the evening of October 28, Unitree Robotics (688836.SH) released its 2026 third-quarter report — its first financial report since listing on the STAR Market on August 19. For the first three quarters, the company posted revenue of RMB 1.817 billion, up 55.71% year over year; total profit of RMB 413 million, up 143.05%; and net profit attributable to shareholders of RMB 387 million, up 270.54%. The third quarter alone, however, showed ‘revenue up, profit down’ — revenue of RMB 665 million, up 69.93% year over year, but net profit of RMB 113 million, down 17.37% year over year and roughly 49.55% below the second quarter’s RMB 224 million.
Structurally, the profit pullback stems mainly from front-loaded R&D and capacity investment: R&D spending for the first three quarters already exceeded the full-year total of last year, and post-IPO capacity expansion and supply-chain stocking further diluted quarterly profit. On the market side, the A-share humanoid-robot sector rallied strongly on the day of the release: Unitree’s share price briefly broke RMB 700 intraday and closed at RMB 695.9, up 8.74%, with a market cap around RMB 281.5 billion — more than 70% above its recent intraday low. The same evening, the company also announced a planned outbound investment through a wholly-owned subsidiary to double down on its industrial footprint. Along the chain, domestic cost reduction in core upstream components such as motors, reducers, and lead screws is forming a positive feedback loop with robot volume — the fundamental support beneath this sector rally.
From an industry perspective, Unitree’s first quarterly report is a sample of the pivotal moment when ‘physical AI shifts from prototype validation to scaled deployment’: 55.71% revenue growth validates the commercialization path from conference-stage star to volume delivery, while the deliberate profit pullback shows top players prioritizing cash flow into R&D and capacity positions — a financial profile typical of growth-stage platform hardware industries, from smartphones to electric vehicles. The RMB 281.5 billion market cap prices not current earnings but the option value of an ’embodied-AI entry point.’ Over the next two to three quarters, the trajectory of deliveries and gross margin will determine whether this valuation holds.
For Junyi Microelectronics, the ramp-up of the humanoid-robot industry is opening a second growth space for edge-computing hardware: motion control, multimodal perception, and on-device inference in robot bodies demand high-compute, low-power controllers and compute boards — demand that shares the same origin as the embedded-compute needs served by the J-3568/J-3576/J-3588 core boards and K-series edge-AI motherboards. Unitree’s R&D intensity — spending above all of last year — signals a long-term demand curve for compute boards and edge control hosts across the robot supply chain, and the 5G-EDGE-3399Pro and AI-BOX-3588 edge-computing hosts can provide the compute carrier for edge-side data loops in robot training and deployment. The IPC-i5-1235U industrial mini-PC and touch-panel all-in-one machines can likewise take on supporting deployments at robot production-line stations and dispatch control centers. Every advance in physical AI creates a new interface for edge-computing hardware.
Source: https://baijiahao.baidu.com/s?id=1880294001547965176 (2026-10-28)
