The artificial intelligence boom has been dominated by what software can say, write, and predict. Now, investors are shifting capital towards machines that can act.

US investors are increasingly backing robotics and other forms of “physical AI” systems that combine software and hardware to automate real-world tasks in factories, warehouses, construction sites, and transportation networks. The shift marks a recalibration of venture priorities as global competition intensifies, and the economic limits of purely digital AI become clearer.

In 1Q26, North American physical AI startups attracted USD 24.2bn in funding and M&A, the highest total on record, according to a review of Mergermarket data. Waymo’s USD 16bn funding round in February, which valued Alphabet’s self-driving car unit at USD 126bn post-money, accounted for the bulk of activity.

Other notable deals included a USD 1.5bn Series G round for drone developer Shield AI, valuing it at USD 12.7bn post money, and robot intelligence systems designer Rhoda AI, which emerged from stealth with USD 450m in Series A funding at a USD 1.7bn valuation.

Source: Mergermarket, data correct as at 14-Apr-26  

Playing catch up

The surge in venture and growth capital reflects a growing conviction that the next wave of value creation will come from applying AI to the physical economy, where productivity gains are larger, switching costs higher, and defensibility stronger, investors said.

At stake is more than venture returns, but economic resilience, noted Emma Norchet, the lead private technology investor at T. Rowe Price, which participated in Physical Intelligence’s USD 600m Series B round late last year. Nations that automate production domestically can control costs and supply chains, while those that cannot risk structural dependence on foreign manufacturing.

The US, long dominant in digital innovation, is behind the curve in automation-intensive fields such as robotics, advanced manufacturing, and electrical engineering, said Ross Diankov, an American who in 2011 went to Japan to found Mujin, which develops software and controllers for industrial robots. Mujin established a US headquarters in Atlanta in 2021 to expand across North America.

“There is finally an understanding of a need for automation here in the States,” he said.

That realization is sharpened by persistent comparisons with China, which has invested heavily in robotics and applied AI, added Tosh Kojima, managing director at DC Advisory.

Hardware as moat

Unlike the previous SaaS cycle, many of today’s most compelling startups are hybrid businesses that integrate advanced AI with bespoke hardware, Kojima noted. Warehouse robots, autonomous vehicles, intelligent manufacturing lines, and energy systems are capital-intensive, operationally complex, and harder to scale. But they are also harder to displace.

Hardware, once considered a venture risk, is increasingly viewed as a moat, according to Kojima. The new investment thesis is less about choosing between hardware and software than about owning the intersection of both, Diankov and Kojima said. Even companies that appear purely digital are expanding into physical domains, training models for robotics, healthcare, and industrial operations, they said. Conversely, hardware companies now depend on sophisticated AI optimization, simulation, and low-level systems engineering. Few serious players operate without deep integration across the stack, Diankov said.

From transport to humanoids

Transportation illustrates this convergence. Blue Water Autonomy is embedding AI into manufacturing to build fully unmanned, ocean‑going vessels designed for military missions. “We’re applying the same techniques used in large learning models to the physical world,” said CEO Rylan Hamilton. “Tasks that were extremely difficult in robotics five years ago are now possible.”

Regent Craft is pursuing a similar approach in aviation, developing electric aircraft that skim above ocean waves, using AI to optimize safety, performance, and energy efficiency.

Strong government and defense demand underscores the strategic value of physical AI platforms for investors. Founded in 2024, Blue Water has raised USD 64m, including a GV‑led Series A, and plans another round. Older startup Regent has secured USD 100m from Hawaiian Airlines, Japan Airlines, Lockheed Martin, and others.

The economic pressure to embrace physical AI is acute. Industries facing labor shortages and physically demanding work, including manufacturing, logistics, construction, energy, and mining, are early adopters.

In construction and mining, some firms anticipate workforce gaps in the tens of thousands over the next decade. FieldAI and Dusty Robotics are among the well-funded startups focused on these job sites, while incumbents like Caterpillar and John Deere are heavily investing in autonomous machinery, precision agriculture, and strategic partnerships with robotic companies.

John Deere, for example, invested in Apptronik’s USD 520m Series A extension in February. Apptronik’s humanoid robot, Apollo, is capable of heavy physical labor.

Tesla is also entering the humanoid race, reallocating factory space from electric vehicles to accelerate production of its Optimus robot. T. Rowe Price’s Norchet expects Tesla to emerge as the dominant player.

Other humanoid developers include Figure AI, which raised over USD 1bn in Series C funding at a USD 39bn valuation in September, and Agility Robotics, which has reportedly raised more than USD 640m since 2015.

Fragmentation and risk

Unlike digital AI, which has consolidated around a handful of dominant platforms, physical AI is highly fragmented. Building deployable systems requires coordination across hardware manufacturers, AI model developers, data infrastructure providers, and application-specific integrators. No single company controls the full stack.

Some investors see opportunity in that fragmentation. They say it creates room for vertical specialization and for startups focused on bottlenecks such as data collection, edge computing, autonomy software, or human-machine interfaces. Others see risk. Integration remains difficult, deployment cycles are long, and failures are costly.

Several advanced robotics startups have already shut down due to high development costs and failed commercialization, including Cartwheel Robotics this year, K-Scale Labs in 2025, and Alphabet’s Everyday Robots in 2023.

Still, momentum is building. Falling compute costs, improving models, and sustained labor and supply-chain pressure are pushing companies toward automation. The remaining challenge may be architectural, according to executives. Today’s AI systems excel at prediction but struggle with physics, causality, and real-world reasoning, they say. Closing that gap will require new designs that tightly integrate perception, action, and feedback.

For investors, the opportunity is clear: physical AI offers the chance to back companies that reshape the productive base of the economy. The rewards may arrive more slowly than their software bets, but they may also prove more durable.