What Physical AI Startups Are—and Why Funding Is Exploding
Physical AI startups are companies that combine artificial intelligence with hardware, robotics, or engineering tools to design, control, or automate complex physical systems in the real world, turning digital models into machines, products, and processes that move, build, and manufacture. After several years of hype around language models, investors are now shifting attention to embodied AI and AI hardware investment that promise direct impact on factories, supply chains, and laboratories. These firms sit at the intersection of software and robotics, translating algorithms into physical output such as faster product design cycles or automated production lines. Unlike pure software, their value depends on how well they compress costly engineering work—simulation, testing, and manufacturing—into AI workflows. That promise is driving mega robotics funding rounds and embodied AI valuation levels that rival, and sometimes exceed, leading large language model players.
Prometheus and the Rise of the Artificial General Engineer
Prometheus, co-founded by Jeff Bezos and Vik Bajaj, has raised USD 12 billion (approx. RM55.2 billion) at a USD 41 billion (approx. RM188.6 billion) valuation, making it one of the most valuable physical AI startups in the world. The company’s goal is an “artificial general engineer” that can automate the design and manufacturing of complex systems, from jet engines to pharmaceutical compounds. A large portion of the capital will go to compute, underlining how hardware-intensive embodied AI has become compared with cloud-only software models. According to CNBC’s interview with Bezos, AI-driven productivity gains will “create labour scarcity rather than unemployment,” framing physical AI as a growth engine for industrial jobs, not a replacement. With about 150 employees across multiple engineering hubs, Prometheus signals that mega-rounds are no longer reserved for language-focused AI giants.
PhysicsX Compresses Months of Engineering into Seconds
PhysicsX, based in London and founded by former Formula 1 engineers, raised USD 300 million (approx. RM1.38 billion) at a USD 2.4 billion (approx. RM11.04 billion) valuation, more than doubling its worth from a prior round near USD 1 billion (approx. RM4.6 billion). The company builds AI-native engineering software that turns complex simulations and design processes—which once took months—into workflows that run in seconds. It serves manufacturing and defence customers struggling with resource and skill bottlenecks as products grow more intricate. PhysicsX says it gives engineers the ability to explore thousands of designs instead of a handful, dramatically widening the search space for better aircraft, chips, engines, and energy systems. Backing from Temasek, Nvidia, Applied Materials, and Siemens shows how strategic investors view physical AI as a way to speed up hardware innovation rather than incremental software productivity.
Why Embodied AI Valuations Now Rival Software-Only Giants
The funding arcs of Prometheus and PhysicsX show that physical AI startups can now command embodied AI valuation levels on par with top-tier software AI companies. Investors see a direct path to tangible ROI: automate an expensive engineering step, and the savings or new revenue show up in unit economics, not only in user engagement metrics. Robotics funding rounds in areas like factory automation, simulation-driven design, and AI-managed manufacturing lines are viewed as infrastructure bets, similar to early cloud or semiconductor investments. Capital-intensive compute needs and hardware installs are no longer red flags; they are seen as necessary moats, making products harder to copy than pure code. As AI-infused machines handle more of the physical economy—from engines to energy systems—backers expect returns tied to large, measurable productivity gains rather than speculative platform plays.
