Humanoid Robots: Physical AI's Next $40T Frontier

Goldman Sachs raised its 2035 humanoid robots market forecast sixfold to $38 billion as physical AI deployments scale from factory pilot to full industrial production.

Goldman Sachs raised its 2035 humanoid robotics forecast sixfold — from $6B to $38B — after AI training advances outpaced projections.
Figure AI commands a $39B private valuation; Tesla has earmarked $20B in 2026 capex for Optimus production and compute.
Per-unit prices have fallen from $150,000-plus in 2023 to $40,000–$60,000 in 2026, with Chinese models entering at $16,000.

Lead

The global humanoid robots industry has crossed from speculative bet to boardroom priority in under two years. Goldman Sachs revised its 2035 market estimate to $38 billion — a sixfold upgrade from its prior $6 billion forecast — citing end-to-end AI physical training as the single advance that most outpaced earlier assumptions. Nvidia CEO Jensen Huang has put the total addressable market for humanoid labor automation at $40 trillion, while Morgan Stanley projects the broader ecosystem reaching $5 trillion by 2050. As of June 2026, deployment contracts with automotive and logistics majors are validating unit economics that investors had struggled to model just eighteen months ago.

What Happened

The investment surge traces to three concurrent forces: commoditized hardware, simulation-based software training, and real-world deployment agreements that anchor the commercial case.

Figure AI, the highest-valued private humanoid firm, raised $1 billion in September 2025 to reach a $39 billion valuation. Its Figure 02 industrial humanoid is being piloted with BMW in automotive assembly; Figure 03, the next-generation platform, adds palm cameras, tactile sensors that detect forces as small as 3 grams, and wireless charging. Tesla (TSLA) is pursuing scale through vertical integration. The company has committed $20 billion in 2026 capital expenditure to cover Optimus manufacturing and the compute infrastructure required for mass deployment. The retail price target for Optimus stands at $20,000–$30,000, with a manufacturing cost goal of $20,000 per unit. Elon Musk has characterized the program as a potential driver of 80% of Tesla's long-term enterprise value. Boston Dynamics has committed its entire 2026 Atlas production run — first deliveries going to Hyundai manufacturing facilities. Agility Robotics signed a Robots-as-a-Service agreement with Toyota Motor Manufacturing Canada following a year-long pilot; seven Digit units now handle material logistics at the Woodstock, Ontario plant that builds the RAV4.

Price Compression and the Adoption Curve

The most consequential tech innovation of the past 24 months in this space is price compression. Commoditized actuator components combined with simulation-trained software have driven commercial-grade humanoid pricing from above $150,000 in 2023 to the $40,000–$60,000 band in 2026.

Chinese manufacturers are pushing the floor significantly lower. Unitree's G1 humanoid carries a $16,000 price tag; AgiBot's A2 is available at roughly $35,000. These figures reflect advantages in microelectronics, precision motors, and rare earth magnets — components where China holds structural supply-chain dominance and where Western competitors remain dependent on Chinese inputs.

Nvidia's Infrastructure Play

Nvidia (NVDA) is positioning itself as the infrastructure layer for physical AI in the same way it did for large language models. At Computex 2026, Jensen Huang unveiled the Isaac GR00T H2+ reference design — a hardware-software blueprint developed with China's Unitree and Singapore's Sharpa. The design standardizes integration decisions across data collection, simulation training, and real-world deployment, lowering the barrier to entry for new manufacturers seeking to build on Nvidia's compute stack.

Huang described the present moment as "the ChatGPT moment for robotics" and projected that widespread adoption is three to five years away. At CES 2026, China-based manufacturers dominated the exhibition floor, underscoring the geographic dimension of the technology race.

Geopolitical Dimension

Beijing has designated specific cities as humanoid robot industrial hubs and is channeling state subsidies to domestic manufacturers. The strategic rationale is demographic as much as economic: China faces a contracting labor force and has framed humanoid robotics as a partial structural solution. Speculative projections suggest eventual deployment of hundreds of millions of units domestically.

Washington has not yet imposed sector-specific controls on humanoid robotics comparable to those governing advanced semiconductors, but the policy environment is under active review. The U.S. robotics industry's hardware dependence on China is a recognized vulnerability that adds regulatory uncertainty to an otherwise bullish investment thesis.

Market Scale and Projections

Goldman Sachs projects global humanoid shipments of 50,000–100,000 units in 2026, scaling to more than 250,000 by 2030, primarily for industrial use. The broader physical AI market — spanning robotics, autonomous vehicles, and AI-enabled industrial systems — was valued at $81.4 billion in 2025 and is projected to reach $1.15 trillion by 2035, compounding at 33.5% annually. SoftBank's Masayoshi Son has publicly identified the sector as the most probable source of the next trillion-dollar company.

Outlook

The humanoid robotics sector is transitioning from venture-funded experimentation to structured industrial deployment, supported by falling unit economics, enterprise-grade contracts, and a platform infrastructure push from the world's most valuable semiconductor company. Goldman's sixfold upward revision and the scale of capital being committed by Tesla, Figure AI, and Nvidia reflect a market consensus that the technology has crossed a commercial viability threshold. Near-term risks include supply chain concentration in China, regulatory uncertainty in Washington, and a pricing race that may compress margins before production volumes justify them. The structural case — decades of industrial and, eventually, consumer labor automation — remains intact and expanding.

Mentioned tickers: TSLA, NVDA