Physical AI, robotics, smart cities: the next global capex cycle

Facts. Humanoid robotics: $8.7B in VC from January to July 2026; market estimated at ~$4.2B in 2026 → ~$40.5B in 2033 (CAGR 38.2%); Goldman Sachs: $38B in 2035 (bull case $152B)[30]; Morgan Stanley: 930 million units in 2050 (China 302.3M, US 77.7M), unit price dropping from $200k (2024) to ~$50k (2050).

Industrialization signals: Unitree IPO in Shanghai ($905M)[26]; Agility Robotics heading toward listing via a $2.5B merger backed by >$300M in orders; Figure 03 returning to BMW Spartanburg after contributing to ~30,000 vehicles in 2025; 1X in full-scale production in California; AGIBOT: 10,000th humanoid produced. The global robotics market exceeds ~$60B, with 590,000 industrial installations per year. Investor perspective.

The pattern is one of a capex cycle: robotic foundation models + labor shortages + tariff-driven reshoring = automation economy becoming irresistible.

Two professional cautions: (1) 2025 deployments were >85% educational/demonstrators, industrial adoption remains unproven in terms of cycle time, energy, and maintenance (IFR); (2) the concentration of value chains (China: 80-90% of 2025 installations, US ban on new foreign robotics) creates a geopolitical supply risk. Smart cities & rent connection: smart cities = fleet buyers (logistics, security, airport ground ops, JAL/GMO in Japan) financed via robot-as-a-service: exactly the type of securitizable contracts from Idea No. 2. Biotech/MANBRIC: same logic of long-term capex financed by private debt, the financial transmission vector is identical.