Robotics Investment
Command Center
Physical AI is the next wave. Track the thesis, monitor positions, and stay ahead of the humanoid robotics revolution.
Market Overview
Key metrics and current state of the robotics sector
Manufacturing costs declined 40% YoY vs earlier projections of 15-20%. Unitree launched R1 humanoid at $5,900 — price point previously thought impossible. Commercial pilots transitioning to operations at BMW, Amazon, GXO.
Tesla targeting 5,000 Optimus units in 2025, scaling to 100,000 by 2026. BYD aims for 1,500 humanoids in 2025, ramping to 20,000 by 2026. Agility Robotics scaling tens of thousands annually.
China recorded 610 robotics investment deals totaling $7B in first 9 months of 2025 (250% YoY increase). Controls 85%+ of rare earth refining. National support for "embodied AI" far exceeds other nations.
Figure AI: $39B valuation with Microsoft, OpenAI, Nvidia backing. Apptronik raised $1B+ in 2025. Physical Intelligence: $400M. Skild: $300M Series A. IPO wave beginning: Unitree, Otsaw, more in 2026.
Investment Thesis
The roadmap from AI to Physical AI dominance
Primary Thesis: Robotics represents the application layer of AI — where foundation models meet the physical world. As AI transitions from digital to physical, the value chain shifts to hardware, sensors, actuators, and materials.
Edge Strategy: Rather than picking OEM winners (high risk), focus on the "picks and shovels" that supply all manufacturers. Compute (NVDA), sensors (Keyence, TXN), motion control (ABB, ROK), and critical materials (MP) win regardless of which robot brands dominate.
Timing: 2025-2026 is the accumulation phase before mainstream recognition. Similar pattern to AI in 2022 before the GPT breakout. Commercial deployments validate the thesis while private market valuations signal smart money conviction.
Watchlist
Tiered investment universe for the robotics thesis
TIER 1: PICKS & SHOVELS (Core Positions)
TIER 2: THEMATIC ETFs (Diversified Exposure)
TIER 3: OPTIONALITY (Speculation/Catalyst Plays)
IPO WATCH (Coming Public)
Education Center
Understanding the robotics value chain and key concepts
The integration of AI models into physical systems. Unlike pure software AI, embodied AI requires real-world perception, physical actuation, and safety systems. This is where AI "meets the atoms."
Key Players: NVIDIA (GR00T), Google DeepMind, Physical Intelligence, Skild AI
Motors, gearboxes, and drive systems that create physical movement. Humanoid robots require high-torque, compact actuators in every joint. Each robot may contain 20-40 precision actuators.
Investment Angle: This is a bottleneck. High-precision gearheads and harmonic drives are supply-constrained.
Camera systems, LiDAR, and depth sensors that allow robots to perceive the environment. Combined with AI, enables object recognition, spatial mapping, and safe navigation.
Key Players: Keyence, Mobileye, Sony (CMOS), Luminar, Onsemi
NdFeB (neodymium) magnets are essential for high-torque motors. Each humanoid contains 2-4kg of rare earth magnets — sometimes more than an EV. China controls 85%+ of refining.
Investment Angle: MP Materials is the only active US rare earth mine. Geopolitical hedge.
Humanoid: Bipedal, human-form robots designed to work in human spaces without infrastructure changes. Higher complexity, broader application.
Cobot: Collaborative robots — typically arms — designed to work alongside humans. Already deployed at scale. Lower complexity, proven ROI.
Near-term (2025-2027): Warehouse logistics, automotive assembly, palletizing, material handling
Medium-term (2028-2032): Healthcare/eldercare, retail, hospitality, agriculture
Long-term (2033+): Consumer/household, construction, general-purpose assistance
| Layer | Function | Key Players | Thesis Fit |
|---|---|---|---|
| Materials | Rare earths, copper, lithium, steel | MP, FCX, SCCO | Bottleneck |
| Components | Motors, actuators, gearboxes, encoders | ABB, ROK, FANUY, Yaskawa | Picks & Shovels |
| Sensors | Vision, LiDAR, force/torque, proximity | Keyence, TXN, ON, MBLY | Picks & Shovels |
| Compute | Training, inference, edge processing | NVDA, AMD, QCOM | Core Position |
| Software | AI models, simulation, fleet management | NVDA (Isaac), Unity, ANSS | Picks & Shovels |
| OEMs | Robot manufacturers/integrators | TSLA, Figure, Agility, UBTECH | High Risk/Reward |
| Deployers | End customers operating robots | AMZN, BMW, GXO, Foxconn | Indirect Beneficiary |
Resources
External links and research sources