physical AI—the combination of robotics, advanced artificial intelligence and sensor-based systems—is entering an early growth phase that could lead to significant industrial disruption. The category includes autonomous machines, smart manufacturing systems and other technologies capable of perceiving and acting in real-world environments.

According to PitchBook, robotics and physical AI startups raised $27.6 billion across 1,009 deals in 2025, more than twice the previous year’s total. The funding surge has drawn attention to companies developing general-purpose robotic systems and foundation models for physical tasks.

Genesis AI, for example, emerged from stealth with a $105 million seed round and demonstrated a wheeled robot designed for commercially relevant environments without attempting to reproduce the human form. Dyna Robotics raised a $120 million Series A roughly a year after its founding for foundation models intended to help robots learn service tasks in hotels, restaurants and laundromats.

Physical AI systems combine multimodal sensors, edge computing, simulation environments, robotics foundation models and world-modeling technologies. Together, these capabilities allow machines to gather information, interpret spatial context, evaluate possible outcomes and take autonomous action while improving through feedback from real-world operations.

The economics of the technology are also changing. Components such as lidar, actuators, depth cameras and other sensors have become less expensive. Waymo said its sixth-generation Driver system reduced costs while increasing resolution, range and computing power. The system uses 13 cameras, four lidars and six radars.

Manufacturing, automotive, logistics and healthcare are among the sectors attracting the most capital and experimentation. Companies are testing robotics to address labor costs, shortages of qualified personnel and the need to operate in demanding environments while increasing productivity.

Figure AI, whose humanoid robots assemble automotive parts in BMW factories, reached a $39 billion valuation within three years of its founding, according to the article. The company’s rise reflects investor expectations that general-purpose robot models could have an impact on hardware comparable to the effect large language models have had on software.

The development of physical AI follows a three-stage pattern. The first is a surge of founders and capital around a new capability. The second is a corporate adoption barrier, when organizations move from pilots to production but encounter procurement delays, integration challenges and internal resistance. The third is industrialization, when a smaller group of companies turns the technology into operational infrastructure.

Physical AI is described as being at the beginning of the first stage. The next challenge will be crossing the adoption barrier. Companies will need to integrate robotic systems into existing operations, adapt processes and demonstrate reliable commercial results rather than simply attract investment.

The article argues that the strongest long-term advantage will come from continuous operational learning. Physical interaction can generate data, which improves simulations and models; those improvements can then enhance performance in real environments. As deployments expand, the resulting infrastructure, data and feedback loops may create advantages that become increasingly difficult for later entrants to replicate.

For corporate leaders, startup funding activity can therefore serve as an early indicator of where future competitive pressure is developing. Rather than viewing deal flow only as a source of potential vendors, organizations can use it to identify emerging technologies and adjust their development and strategic plans before the market reaches large-scale industrialization.

Sources: The Robot Report.