Humanoid Robots on the Eve of Mass Production: Tesla's Optimus Line Takes Shape, Unitree's IPO Ignites China's A-Share Market

The summer of 2026 sees the humanoid robot sector undergoing a critical shift from conceptual hype to tangible mass production. On one side, Tesla's Optimus production line is physically taking shape at its Fremont factory, even as Elon Musk pours cold water on expectations by stating "initial production will be extremely slow." On the other, Chinese player Unitree Technology has secured its IPO approval for the STAR Market, and UBTECH's consumer robot orders have broken the 10,000-unit mark, suddenly accelerating the capitalization of the industrial chain. The race between US and Chinese tech giants in humanoid robotics is reshaping the global supply chain landscape.

Production Line Takes Shape Amid Cautious Expectations: The "Slow Ramp-Up" Logic of Optimus

On July 1, Elon Musk posted a photo on X with the Optimus mass production team at the Fremont factory, captioned "Visited the Optimus robot production line." The image quickly ignited market sentiment. The vehicle production line, originally used for the Model S and Model X, has been transformed into the first automated line for the Optimus Gen3, boasting a designed annual capacity of up to 1 million units, with plans for approximately 40 additional sub-lines.

The market initially interpreted this as an accelerated timeline for large-scale production, but Musk quickly clarified: "No, Optimus production will be extremely slow at first because everything is new. It's not like building a car." Behind this statement lies the hard reality of industrial constraints.

A single Optimus unit comprises roughly 10,000 new, independent components, including micro-actuators, high-precision reducers, and multi-degree-of-freedom dexterous hands—precision assemblies for which almost no mature, universal supply chain exists. In contrast, while a Tesla electric vehicle has 30,000 parts, the vast majority rely on a decades-old, mature automotive supply chain. The core joints of a humanoid robot face an "impossible triangle" of performance, cost, and reliability. Micron-level assembly requirements drastically reduce initial yields, and a capacity shortfall in any single, tiny component can bottleneck the entire output.

Looking at the production ramp-up, Tesla has set a clear path for Optimus: dozens of units per week in June, 100 to 150 per week in July, around 300 per week in August, and scaling to 1,000 units per week by September. However, the Fremont line is designated for early-stage, small-batch iteration. The robots produced in 2026 will only be used for internal factory testing and data collection, not for external commercial use. True mass production will wait until a new, second-generation dedicated factory in Texas comes online in 2027. The core task at this stage is to refine hardware flaws and perfect motion control algorithms, not to chase shipment volumes.

Unitree's IPO Ignites a Rally: Valuation Anchor Meets Profitability Concerns

On July 2, the China Securities Regulatory Commission (CSRC) officially approved Unitree Technology's IPO registration for the STAR Market. Market expectations now anchor its post-IPO valuation above 100 billion yuan (approximately $14.7 billion), far exceeding the previous expectation of 42 billion yuan (approximately $6.2 billion). On the news, China's A-share robotics sector staged a "Unitree rally" on July 3, with over 40 stocks, including Leaderdrive and Tuopu Group, hitting their daily trading limits and the turnover rate for robotics ETFs soaring. In Hong Kong, newly listed Laifual Harmonic (03952.HK) surged nearly 12% in a single day, while leading companies like UBTECH (09880.HK) and Yuejiang (02432.HK) also briefly rallied over 10%.

However, Unitree Technology's latest prospectus reveals a mixed picture. In the first quarter of 2026, the company's revenue was 423 million yuan (approximately $62.3 million), with year-over-year growth plummeting to 68.49% from 332.64% for the full year of 2025. Its non-recurring net profit fell to 40.25 million yuan (approximately $5.9 million), a year-over-year decline of 52.55%. The company placed "revenue growth deceleration and operating performance fluctuations" at the top of all its special risk warnings.

In its prospectus, Unitree explicitly warned that Tesla, with its advantages in large-scale mass production, supply chain integration, AI technology resources, and its own factory deployment, will pose direct competition now that the Optimus Gen-3 has announced the start of small-batch trial production. The cost dimension is even more critical—the bill of materials for the Optimus Gen3 has been driven down to $28,000 (~198,000 yuan), approaching the current average unit price of 167,600 yuan (approximately $24,700) for Unitree's humanoid robots. Musk has predicted that when annual production exceeds 1 million units, the production cost will fall below $20,000.

Domestic competition in China is equally fierce. According to GGII, more than 10 automakers have explicitly laid out plans for humanoid robots, including XPeng, BYD, GAC, Li Auto, and NIO. The first card played by automakers is cost control—BYD's seventh-generation prototype has a mass production target price locked within 200,000 yuan (approximately $29,500), and Geely plans to reduce costs to under 150,000 yuan (approximately $22,100) by 2027.

Facing pressure from both sides, Unitree Technology has already initiated a price-cutting strategy. From 2023 to September 2025, the average unit price of its humanoid robots plummeted from 593,400 yuan (approximately $87,500) to 167,600 yuan, a drop of over 70%. However, its gross margin contracted from 87.67% to 62.91% over the same period, with the price decline significantly outpacing cost reductions. The prospectus shows that direct material costs have consistently accounted for a high 72% to 81% of revenue, leaving very limited room for traditional manufacturing cost-down strategies.

Industry Chain Map: Components First, Full Robots Await Validation

Judging by the distribution of stocks hitting their daily limits, component companies in reducers and actuators performed most prominently. Zhongda Leader specializes in reducers, while Wolong Electric focuses on motor systems—sub-sectors with high technical barriers and scarcity. Compared to full-robot manufacturers, whose performance realization still requires time, the investment certainty in the component supply chain is stronger.

International giants are also accelerating their entry. Nvidia announced on June 29 that its robotics team is simultaneously expanding hiring in Beijing, Shanghai, and Shenzhen across four major areas: embodied intelligence, simulation, deployment, and solution architecture. The goal is to connect with China's massive full-robot industrial chain and optimize the technical link from simulation to physical implementation. This signals a potential acceleration in breaking through bottlenecks in AI control and motion algorithms for humanoid robots.

Policy support is also intensifying. China's Ministry of Industry and Information Technology is accelerating the construction of a standards system for the entire humanoid robot industry chain, while cities like Shenzhen, Hefei, and Shanghai are rolling out supporting policies for real-world training and financial subsidies. According to customs statistics, in the first five months of this year, China exported a total of 10.377 million robots, with an export value of 19.99 billion yuan (approximately $2.9 billion), reaching over 150 countries and regions worldwide. Among these, cleaning robot exports reached 14 billion yuan (approximately $2.1 billion), accounting for over 70% of the total.

Investment Logic for the "Year of Mass Production": Supply Chain in the Short Term, Leaders in the Long Term

The industry has defined 2026 as the "first year of mass production" for humanoid robots, but the true meaning of "mass production" needs to be unpacked. The core industry question has shifted from "can we build it?" to "can we build it at low cost, sustainably, and at scale?" However, large-scale commercialization still needs to overcome three major hurdles: technology, cost, and application scenarios.

Tesla plans to prioritize deploying Optimus in its own factories in 2026, accumulating data and experience in a controlled environment before expanding to external markets. Industrial scenarios demand high reliability, home scenarios are cost-sensitive, and service scenarios require complex interaction capabilities—each direction needs time for validation. Tesla expects capital expenditures to exceed $25 billion in 2026, primarily for AI infrastructure and Optimus, but the commercialization of Optimus will take at least 5 to 10 years.

For investors, the time scale needs adjustment: look for opportunities in the supply chain in the short term, small-batch deployment in specific scenarios in the medium term, and leading companies with core technologies in the long term. Sinolink Securities stated that 2026 is a crucial node for the humanoid robot's 0-to-1 realization, with Tesla's supply chain expected to officially begin large-scale mass production, and the shipment volume of leading domestic full-robot manufacturers in China expected to leap from thousands to tens of thousands of units.

Over the past week, more than a dozen robotics ETFs have seen net capital inflows. Data from Hithink RoyalFlush iFinD shows that the E Fund Robotics ETF (159530) alone received a net inflow of approximately 830 million yuan (approximately $122.3 million) last week. Multiple institutions believe that catalysts for the robotics sector will continue to emerge, and the advancement of commercialization will drive both valuation repair and earnings growth for the sector.