Humanoid robots, long a staple of science fiction, are inching closer to reality. Designed to mimic human appearance and movement, these robots are emerging in an environment rife with both hype and technical hurdles. The rush to integrate artificial intelligence (AI) into mobile, human-like forms shows promise for reinventing industries like manufacturing, healthcare, and home services. However, progress remains uneven, and many of the aspirations tied to humanoid robotics are far from being realized.

Humanoid robots in the spotlight

The robotics industry is buzzing with investments, prototypes, and pilot projects. Companies like 1X, Tesla, and Nvidia are betting big on the potential of humanoid robots. For example, 1X Robotics is positioning its prototype, Neo, as a home assistant capable of performing tasks like laundry. As explained by Bernt Børnic, CEO of 1X, Neo relies on a combination of AI models, sensors, and actuators to interpret and interact with the physical world in real time. Unlike traditional industrial robots, which perform pre-programmed tasks in repetitive settings, humanoid robots aim to adapt to unpredictable, multi-purpose environments. This adaptability is powered by advances in “physical AI,” which integrates cameras, sensors, and neural networks into hardware capable of learning over time.

Companies outside 1X are also pursuing this vision. Tesla’s humanoid robot, Optimus, has been discussed by Elon Musk as the next big consumer product, although promises of $30 trillion in annual revenue are almost certainly exaggerated. Nvidia, meanwhile, is developing the AI hardware and software infrastructure required for these humanoid models, underscoring its pivotal role in the AI robotics supply chain.

The promises: Labor transformation and huge markets

Proponents of humanoid robots argue that they could address chronic labor shortages, particularly in manufacturing. The U.S., for example, is projected to face a shortfall of nearly two million skilled manufacturing jobs by 2033. Humanoids are seen as potential solutions for roles that are hard to fill—not just in factories but also in warehouses, hospitals, or even homes. For instance, some companies are piloting humanoids in freezer environments, keeping human workers out of extreme cold while accomplishing essential tasks under remote human guidance.

The market projections are similarly ambitious. Some estimates suggest that humanoid robots, and the broader field of physical AI, could become a multi-trillion dollar industry by 2050. As of 2025, robotics investment and development efforts are breaking new records, with AI-induced enthusiasm playing a strong role in this rise.

While these claims excite stakeholders, skeptics point out that humanoid robots are currently far from viable for mass deployment. The multi-faceted challenges of development highlight the gap between current capabilities and lofty industry promises.

Data: The bottleneck of physical AI

A defining challenge for humanoid robotics is data—specifically, the lack of it. While large language models like ChatGPT revolutionized static, text-based AI, humanoids require far more complex inputs to navigate and operate in physical spaces. Robots need massive datasets to learn, practice, and refine their methods. These include precise recordings of human movements, object manipulation tasks, and sensory inputs.

One way to fill this “robot data gap” is through simulation: creating virtual worlds where robots can experiment and learn at scale. Another approach, known as teleoperation, involves robots being remotely controlled by humans, who provide real-time motor and motion commands. However, these methods have limitations. Synthetic environments often miss the nuances of real-world physics, while teleoperated training is slow, expensive, and labor-intensive.

Ken Goldberg, a professor and roboticist at UC Berkeley, suggests a “flywheel” approach, where robots continuously collect data while performing tasks, using this to improve their neural models iteratively. While this method shows promise, it’s still in its infancy, and widespread deployment of robots in dynamic environments remains a long-term goal.

The hype: Fuelled by AI breakthroughs

Much of the excitement surrounding humanoid robotics stems from general advances in artificial intelligence. Generative AI technologies such as large language models and image-generation systems have set a precedent for the economic and creative potential of data-driven algorithms. Many in the robotics field hope that their “GPT moment” will come, where physical AI experiences a similarly transformative breakthrough. For now, this remains speculative. Creating movement and dexterity in real-world conditions involves deeper complexities than generating text or images on screen.

Tech leaders amplify this enthusiasm, giving rise to what some experts derisively call “humanoid hype.” Figures like Elon Musk and Jensen Huang (CEO of Nvidia) continue to push bold visions for humanoid applications, fueling both public interest and investor dollars. China has also embraced the humanoid robotics race with vigor, announcing $140 billion in funding for robotics and AI in 2025 as part of its broader development strategy.

Engineering challenges and unfulfilled promises

Despite the investments and enthusiasm, humanoid robots face significant technological hurdles. They struggle with core functionalities that humans take for granted: picking up objects, walking on uneven terrain, and working for prolonged periods without recharging. Battery life, durability, and the intricate dexterity needed to manipulate objects all remain critical weak points.

Even pilot programs conducted in controlled environments reveal these limitations. Companies testing humanoids in warehouses report that robots are often too slow, unreliable, or error-prone to be fully trusted with production goals. And while they excel in specialized scenarios like “freezer work,” extending this utility across more complex environments like homes and hospitals is an as-yet unproven challenge.

China’s rapid expansion vs. global competition

China is surging ahead as a leader in robotics manufacturing, benefiting from its well-integrated supply chains. With over 140 companies producing humanoids by 2025, the nation’s scale and efficiency have set it apart globally. Robots from firms such as UBTech and Unitree are already being deployed ahead of competitors, reinforcing China’s position as the dominant supplier of industrial robots.

However, long-term success will depend less on quantity of deployment and more on qualitative advances in AI integration and real-world applicability. As with other “Made in China” strategies, overproduction could potentially lead to bubbles without delivering the functional breakthroughs that the robotics market needs.

The road ahead

Humanoid robots have the potential to transform industries and address global challenges, but these advances will not come overnight. Current progress—whether through better AI, innovative engineering, or scaling manufacturing—faces natural limits in development time, resource costs, and social adaptation periods. For all their potential, today’s humanoids still wobble in their quest to move beyond the laboratory.

If humanoid robots are ever to fulfill their promise, their developers will need to overcome hurdles in data collection, hardware precision, and energy efficiency. At the same time, broader societal acceptance—and regulation—will play a key role in shaping their rollout. Until then, the gap between vision and practicality may remain vast, leaving humanoid robots more compelling on the conference stage than on an assembly line.