Morgan Stanley’s "Humanoid Tech 25": Betting on the Components, Not the Brands

The projection is staggering, if distant: a global market worth $5 trillion by 2050, with over one billion humanoid units in operation. But according to a new analysis from Morgan Stanley, the smartest way to play this robotic future isn't to bet on the robots themselves.

In a comprehensive report released this month, the bank unveiled its "Humanoid Tech 25"—a list of companies poised to dominate the sector not by building the machines, but by supplying their guts, eyes, and brains.

The report advocates for a classic "picks and shovels" strategy. With over 150 humanoid startups currently operating in China alone—a density that Beijing has warned is creating a "bubble"—picking the ultimate winner among OEMs (Original Equipment Manufacturers) is a high-risk gamble.

Instead, Morgan Stanley argues that whether the future is dominated by Tesla, Unitree, or Figure, the machines will all require the same underlying infrastructure: specialized sensors, power management chips, and massive compute capabilities.

The "Picks and Shovels" Ecosystem

The bank breaks the opportunity down into three distinct "buckets," revealing a supply chain that is heavily weighted toward Asian hardware giants and specialized component makers rather than the American software firms that often dominate AI headlines.

1. The Brains (AI, Compute & Design)

While NVIDIA, AMD, and ARM are the expected heavyweights for onboard compute—necessary for processing visual data and maintaining balance in real-time—the report highlights a less obvious sector: Electronic Design Automation (EDA).

As humanoid robots evolve, they will increasingly require specialized Application-Specific Integrated Circuits (ASICs) rather than off-the-shelf chips. This shifts leverage to firms like Synopsys and Cadence, which provide the software architecture used to design these custom silicon brains.

On the software side, the inclusion of Chinese giants Baidu, iFlytek, and Alibaba acknowledges the geopolitical reality of the market. While the West relies on OpenAI or Google, China’s domestic robot army will run on domestic Large Language Models (LLMs) and Vision-Language-Action (VLA) systems.

2. The Eyes & Senses (Perception)

Humanoid robots require a density of sensors far exceeding that of autonomous vehicles, often needing six or more cameras per unit alongside depth sensors.

• Sony: Identified as a volume beneficiary due to its dominance in CMOS image sensors.
• Hesai: A leader in LiDAR. The report notes a divergence here: while Tesla famously rejects LiDAR, most industrial humanoids (like those from Xiaomi or XPeng) rely on it for precise mapping and safety in factory environments.
• Melexis: Perhaps the most specific technical call in the report, this Belgian firm is highlighted for its Tactaxis technology. Using magnetic sensors, this allows robots to develop a "sense of touch" delicate enough to handle fragile objects—a possibly critical hurdle for moving robots from factories to homes. The firm also specializes in stray-field-immune sensors, which are vital when packing sensitive electronics next to the high-interference magnetic fields of electric motors.

3. The Body (Actuation & Power)

The report identifies a massive pivot occurring among traditional automotive suppliers. Companies like Joyson Electronics and Desay SV are retrofitting manufacturing lines designed for car parts to mass-produce robot actuators.

Similarly, Samsung (including both its Electronics and Electro-Mechanics divisions) is aggressively targeting the actuator and battery markets. The company recently confirmed its entry into the humanoid race with a distinct "provider and customer" strategy, planning to use the machines in its own manufacturing lines while selling the underlying technology to others. To support this component ecosystem, Samsung Electro-Mechanics has invested in Norwegian motor specialist Alva Industries, securing access to "FiberPrinting" technology—a method for creating lightweight, ironless motors critical for next-generation robotic hands.

The list also leans heavily on legacy chipmakers—Texas Instruments, STMicroelectronics, Infineon, and Renesas—which produce the Motor Control MCUs and Power Management ICs (PMICs) required to regulate the 20–40 motors found in a typical humanoid.

The $200,000 Barrier

Despite the $5 trillion forecast, Morgan Stanley offers a sobering reality check on the near-term timeline. The bank predicts adoption will remain "relatively slow" until 2035, primarily due to cost.

Currently, a capable humanoid robot costs roughly $200,000 to produce. The survey data corroborates this friction; as we reported regarding Morgan Stanley's survey of Chinese executives, 92% of potential buyers indicated that unit costs must fall below $28,000 (200,000 yuan) for widespread adoption to become feasible.

The bank estimates the price needs to drop to $50,000 to trigger the first wave of mass adoption, a target that puts immense pressure on the very supply chain identified in the "Tech 25" list to lower Bill of Materials (BOM) costs.

The Vertical Integration Risk

The report's optimistic view of suppliers does face one existential threat: Vertical Integration.

The primary competitor to these component makers is not another supplier, but the customers themselves. Tesla, for instance, is famously attempting to build its own actuators and sensors for Optimus to strip out cost. If the largest players succeed in bringing component manufacturing in-house, suppliers like Joyson or Hesai could lose their biggest potential revenue streams.

However, the "Tech 25" list suggests that Morgan Stanley believes the market will be too fragmented for full vertical integration to become the norm. With governments in the U.S. and Japan pushing for national supply chain alliances, the industry appears to be moving toward a collaborative ecosystem rather than a "winner-take-all" manufacturing model.

The Humanoid Tech 25

The key companies identified by Morgan Stanley as the backbone of the humanoid era:

• AI & Compute: NVIDIA, AMD, ARM, Synopsys, Cadence, Baidu, iFlytek, Alibaba.
• Sensors & Vision: Sony, Hesai, Melexis, Ambarella, Horizon Robotics.
• Actuation & Power: Samsung Electronics, Samsung Electro-Mechanics, Texas Instruments, STMicroelectronics, Infineon, NXP, Renesas, Onsemi, ROHM, Microchip.
• Auto-to-Bot Pivots: Joyson Electronics, Desay SV.