Humanoid AI: A Coming Wave for Rare Earth Demand

Highlights

• Longbing Cao's landmark review reveals humanoid robotics could reach $38 billion by 2035, driven by AI and generative technologies.
• Humanoid robots depend critically on rare earth permanent magnets and oxides, creating potential supply chain challenges.
• The emerging humanoid robotics market represents a new structural demand layer for rare earth materials beyond existing sectors like EVs and renewables.

---

Longbing Cao (opens in a new tab) of Macquarie University last year authored a landmark review, AI Robots and Humanoid AI: Review, Perspectives and Directions (opens in a new tab) (ACM Computing Surveys, 2024). The paper traces six decades of humanoid robotics and highlights how the integration of generative AI, large language models (LLMs), and large multimodal models (LMMs) is transforming humanoid robots from mechanical curiosities into interactive, real-time platforms for service, manufacturing, healthcare, and defense. Market projections suggest humanoid robotics could reach USD $13.8 billion by 2028, expanding to USD $38 billion by 2035.

Implications for Rare Earth Supply Chains

The connection to rare earths is direct and urgent. Drones, humanoids, and robotic systems rely on rare earth permanent magnets (NdFeB, SmCo) and rare earth oxides (dysprosium, terbium, praseodymium) to power motors, sensors, actuators, and precision guidance systems. The projected growth of humanoids overlaps with booming demand from drones, electric vehicles, and industrial robotics—sectors already straining supply. If humanoid robotics follows the trajectory outlined by Cao, magnet and oxide demand could accelerate well beyond EV-driven forecasts, forcing policymakers and investors to reassess capacity requirements.

Longbing Cao of Macquarie University

Strengths of the Study

The review excels in its integration of AI trends with robotics, painting a credible picture of how generative AI and robotics converge into new commercial categories. The use of market projections from both Goldman Sachs and MarketsandMarkets lends economic weight to the argument, and the taxonomy of humanoid applications—from caregiving robots to industrial humanoids—helps quantify breadth of demand.

Limitations and Blind Spots

The study is less focused on the raw material underpinnings of this robotics boom. While the review carefully maps out software and AI advances, it does not consider the physical constraints: the scalability of high-torque electric motors, the supply of rare earth oxides for high-temperature magnets, or the geopolitical choke points in refining capacity. Furthermore, market projections carry inherent uncertainty. The 50% CAGR forecast risks optimism bias, particularly given that many humanoid prototypes remain proof-of-concept rather than mass-market devices.

Any number of assumptions, factors or events could lead to less robust and optimistic outcomes.

Conclusion

Cao’s review underscores the dawn of humanoid AI as a transformative economic force. For the rare earth sector, this means new layers of structural demand on top of EVs, renewables, and defense technologies. Investors and policymakers must factor humanoid robotics into long-range demand scenarios. The challenge is not only technological but geopolitical: without diversified rare earth supply chains, humanoid adoption could deepen Western dependence on China’s magnet ecosystem.

Citation: Longbing Cao. 2024. AI Robots and Humanoid AI: Review, Perspectives and Directions. ACM Computing Surveys.

©!-- /wp:paragraph -->