Report Interpretation

Overview

J.P. Morgan initiates coverage of Unitree with an Underweight rating despite a constructive long-term view of humanoid and legged robotics. The central argument is that Unitree has a capable, cost-efficient hardware platform, but its current valuation assumes durable leadership while competition, ASP deflation, regulation and the migration of value toward manipulation, AI models and data threaten that premise.

Core views

J.P. Morgan expects humanoid and legged robotics to reach a commercialization inflection in two to four years as technology matures and adoption broadens. It forecasts global humanoid demand rising from 60k units in 2026E to 1.75m in 2030E, with China increasing from 50k to 900k units. The report argues that deployment—not demonstrations—is now the decisive test: repeat enterprise orders, measurable customer ROI, reliability and margin durability must replace scarcity-driven enthusiasm. Structured applications such as inspection, guided delivery, sorting and manufacturing support are viewed as credible early use cases. Unitree is presented as a leading Chinese hardware platform with strengths in vertical integration, motion control, rapid iteration, cost-down execution and a broad product portfolio. Revenue rose from Rmb123m in 2022 to Rmb1.7bn in 2025, while the company has generated positive operating cash flow since 2023 and ended 2025 with Rmb1.4bn of cash. Yet J.P. Morgan says the investment debate is moving from volume growth to earnings quality. It forecasts total sales above Rmb5bn in 2028E, a 46% CAGR from 2025-28, but expects growth to slow as competition and pricing intensify. The report's principal concern is valuation. Unitree traded at roughly 46x FY27E P/S, which J.P. Morgan considers too demanding for a business exposed to lower prices, slower growth and an uncertain ability to capture software and data economics. The Rmb300 Dec-27 target uses 30x FY27E P/S, benchmarked against high-quality global comparables and recent Agility Robotics fundraising. J.P. Morgan uses sales rather than earnings because revenue, shipments and ASPs are more informative at this early commercialization stage, while profitability remains highly sensitive to adoption timing, warranty and service costs, ASP normalization and operating-expense scaling. Competition is expected to erode both pricing power and share. Unitree's average humanoid ASP declined from Rmb593k in 2023 to Rmb260k in 2024 and Rmb166k in 2025; the report forecasts a further 35% blended ASP drop in 2026 and a decline to Rmb72k in 2028E. Its China humanoid share is forecast to drop from 35% in 2025 to 15% in 2028 and 8% in 2030 as local supply chains mature and lower-cost competitors enter. Blended gross margin is forecast to contract by 2-3 percentage points year-on-year through 2028, even though Unitree's manufacturing integration supports lower unit costs. Enterprise deployments could also carry installation, maintenance, training, fleet-management and uptime-support costs that make product-level gross margins less representative of true economics. J.P. Morgan argues that the industry is shifting from robot bodies and locomotion toward dexterous manipulation, embodied-AI models, real-world data and vertical applications. Unitree's core strength remains locomotion and affordable robot bodies, but the report sees limited evidence of a leading in-house physical-AI model capability. In NVIDIA's Isaac GR00T reference platform, Unitree supplies the H2 Plus body while tactile hands and compute contribute differentiated functionality. The report therefore warns that Unitree could remain a successful hardware supplier without receiving the valuation associated with leadership in the intelligence layer. Regulation is another near-term overhang. The report says recent US FCC restrictions on foreign-produced advanced robotic devices may constrain access for Chinese humanoid and quadruped OEMs. Overseas revenue represented 43% of Unitree's 2025 sales, with the US about one-third of that amount; the company also faced additional policy uncertainty after its addition to the US Department of Defense Section 1260H list in June 2026. J.P. Morgan believes these factors reduce international-growth upside and add uncertainty around future launches, although non-US demand could partly offset US constraints. The report also identifies material upside scenarios. Faster conversion of pilots into repeat enterprise fleets, successful hardware-software and data-platform integration, strategic AI partnerships, sustained difficulty in scaling reliable robot bodies, or geopolitical de-escalation could improve Unitree's platform value and support a higher multiple. Its base case forecasts Unitree humanoid shipments rising from 5.2k in 2025 to 74k in 2030, but below overall market growth; evidence that customers renew, scale fleet deployments and achieve measurable ROI would challenge that assumption.

Analysis framework

J.P. Morgan combines a commercialization and supply-demand view of humanoid robotics with Unitree's operating forecasts, pricing and margin assumptions, competitive comparisons, balance-sheet analysis and regulatory assessment. It values the company using forward P/S because revenue, shipments and ASPs are considered more meaningful than mature earnings at this stage, then tests the valuation against global robotics benchmarks and the evidence needed for repeat enterprise adoption.

Methodology notes

• Valuation methodsPS valuation

  Forward price-to-sales valuation

  The report applies 30x FY27E P/S to derive its Rmb300 target because Unitree's revenue, shipment ramp and ASP trajectory are viewed as more meaningful than earnings for an early-stage robotics business.

• Industry AnalysisSupply-demand framework

  Humanoid-robot demand, shipment and capacity analysis

  The report forecasts global and China demand, Unitree shipments, market share and capacity to assess whether growth is constrained by demand or manufacturing supply.

• Industry AnalysisVolume-price decomposition

  Shipment growth and ASP deflation analysis

  Revenue forecasts are assessed through projected shipment increases and declining blended ASPs, showing how lower prices can expand adoption while pressuring margins and valuation.

• Competition & strategyValue chain analysis

  Shift from robot bodies and locomotion toward manipulation, software and data

  The report evaluates where value may accrue in physical AI and argues that Unitree must show that it can capture value beyond hardware manufacturing.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Unitree Robotics - A (688836.SS, 688836 CH)

  Primary covered company; a leading Chinese humanoid and quadruped robotics hardware platform facing valuation and commercialization scrutiny.

  Strengths

  Vertical integration, cost-down capability, motion control, broad product portfolio, positive operating cash flow and net cash position.

  Weaknesses

  Limited evidence of leadership in physical-AI models, a hardware-centric value proposition and dependence on declining ASPs.

  Comparison

  Agility Robotics is used as a directional valuation benchmark; the report contrasts Unitree's cost and platform strengths with Agility's enterprise-workflow focus.

  Risks

  ASP deflation, declining market share, margin normalization, delayed enterprise adoption, US policy restrictions and failure to monetize data or software.

• Agility Robotics

  Comparable and directional valuation benchmark for early-stage hardware-centric humanoid robotics.

  Strengths

  Enterprise-workflow focus and a reported c.US$300m multi-year Digit 5 order pipeline.

  Weaknesses

  Valuation comparison depends on revenue recognition and backlog quality.

  Comparison

  J.P. Morgan says Unitree is stronger in cost-down and accessibility, whereas Agility is more focused on enterprise workflows and measurable ROI.

Key data

• Rating and target priceUnderweight; Rmb300 Dec-27 PTTarget implies ~35% downside from Rmb459.65 on 28 September 2026.
• Current valuation~46x FY27E P/SJ.P. Morgan considers the multiple rich; target uses 30x FY27E P/S.
• Humanoid ASPRmb593k in 2023; Rmb166k in 2025Forecast to decline 35% year-on-year in 2026 and to Rmb72k in 2028E.
• China humanoid market share35% in 2025; 15% in 2028E; 8% in 2030EForecast decline reflects stronger competition and a shift toward manipulation and application-layer capabilities.
• Global humanoid demand60k units in 2026E to 1.75m in 2030EChina demand is forecast to rise from 50k to 900k units.
• Unitree revenueRmb1.7bn in 2025; above Rmb5bn in 2028EThe report forecasts a 46% CAGR over 2025-28.
• Margin outlookBlended GPM down 2-3 percentage points year-on-year through 2028Cost-down is expected to be offset by ASP pressure and more complex deployment requirements.

Impact & implications

The report sees a favorable long-term industry opportunity but believes Unitree must prove that its hardware advantages translate into repeatable enterprise economics and a defensible role in data and intelligence. Until then, J.P. Morgan expects investor attention to remain on ASPs, margins, market-share retention, enterprise order conversion and international market access rather than early shipment leadership.

Risks

• Faster-than-expected conversion of pilot deployments into repeat enterprise fleets could improve Unitree's growth and valuation.
• A successful evolution into a hardware-data platform through AI partnerships could strengthen its role in physical AI.
• Reliable, scalable robot bodies may remain harder to replicate than expected, preserving Unitree's hardware moat.
• Geopolitical de-escalation or resilient non-US demand could restore international growth optionality.

What to watch

• Repeat enterprise orders, fleet expansion, customer renewals and measurable ROI.
• Humanoid ASP trends, gross-margin resilience and Unitree's ability to defend market share.
• Evidence of hardware-software coupling, data ownership or monetization, and physical-AI partnerships.
• US FCC implementation, product approvals and the trajectory of overseas demand.
• Progress in manipulation, reliability, safety and other bottlenecks needed for broad enterprise deployment.