Physical AI raises $47.4B in H1, topping 2022-2024
The next AI check is not for another chatbot. It is for the machines that drive, fly, and shoot.
Physical AI companies raised $47.4 billion across 521 deals in the first half of 2026, Crunchbase News reported Tuesday — nearly 80 percent more than the $26.4 billion they took in during the first half of 2025, and more than the $41.9 billion the category raised across 2022, 2023, and 2024 combined. Against the second half of last year the jump is even sharper: almost four times the $12 billion that landed in those six months. Crunchbase's bucket is wide on purpose. It folds in robotics, autonomous vehicles, aerospace, drones, industrial automation, and sensors — anything that puts a model in contact with the physical world.
Read the constituent deals and the boom looks less like a thousand robot startups and more like a barbell. Waymo's $16 billion Series D in February, priced at a $126 billion valuation and co-led by Alphabet, Dragoneer Investment Group, DST Global, and Sequoia Capital, accounts for nearly a third of every venture dollar in the tally. Anduril added $5 billion in May at $61 billion, double the valuation it carried less than a year earlier. Shield AI closed a $2 billion Series G in March at $12.7 billion. Saronic, which builds autonomous sea vessels, raised $1.75 billion the same month at $9.25 billion. Strip those four checks out and the "physical AI wave" shrinks into something closer to a normal up-cycle. The exits tell the same story. Crunchbase says activity concentrated in aerospace, defense, and drones — SpaceX's June IPO, HawkEye 360, Aevex — not in warehouse arms or humanoid demos. The robotics exception was Mobileye's roughly $900 million purchase of Tel Aviv's Mentee Robotics, which Mobileye tied explicitly to a physical-AI push.
That composition is the point. Firms that made their names on software and social media are now writing hardware-sized checks, and they are writing them to companies that already look like defense primes or robotaxi utilities, not lab projects. We mapped the other side of this money yesterday — AI's new kingmakers: chip giants poured $250B into startups in 2026. Today's number is where those checks are landing. Edison Partners' Ryan Ziegler told Crunchbase the interesting shift is not more humanoid rounds. It is cheaper sensors, better simulation, and hardware that behaves like a distribution channel for software and data. "Hardware [as] the distribution model for creating a data intelligence flywheel," he said. Eclipse's Joe Fath put it more bluntly: customers buy reliability and revenue, "not technical sophistication alone." The China data layer is making the same bet from the other direction — we covered that in China Telecom leads funding for embodied-AI data startup Mifeng.
The take is unromantic. Physical AI is winning the allocation meeting because autonomy and weapons platforms already have buyers, multi-year contracts, and valuations that look like software companies. The go-kart clips and fold-up home robots are real, and they are not where $47.4 billion goes. If the next six months fill in beneath the megadeals — factories, farms, utilities, the boring infrastructure Ziegler actually wants — the category is a new industry. If it stays four logos and a headline, it is just last year's software money wearing a helmet.
What to watch: whether H2 funding broadens past robotaxis and defense, or Waymo-and-Anduril keep doing most of the arithmetic.
Is physical AI a new industry, or just four megadeals wearing a category label? Tell us in the comments.
