Robotics Startup Funding Hits $9.9 Billion in Q3, Up 185% for 2026

By Grace Kim

Robotics startups pulled in $9.9 billion in global venture funding in Q3, more than double the $4.3 billion they raised in Q3 2025, according to Crunchbase data. That caps a run of quarters at or near $10 billion, and puts sector funding for Q1 through Q3 2026 up roughly 185% year over year.

The analysis excludes defense-related companies. The money is concentrating fast: China and the U.S. account for the bulk of it, and the unicorn board is growing at a pace the sector has never seen.

Who's getting the money?

China leads with $13.2 billion invested in robotics startups this year. The U.S. follows at $12.1 billion. Germany ranks third but trails far behind at $1.7 billion.

China's largest rounds went to two companies:

• Xpeng Robotics, a Guangzhou-based builder of AI models and physical robots, raised $900 million.
• AI² Robotics, a Shenzhen company building intelligent robots for manufacturing, healthcare and services, closed a $736 million Series C.

The U.S. produced its own billion-dollar rounds. Travis Kalanick's Los Angeles-based Atoms — which spans a ghost kitchen business plus robotics divisions for mining and transport — raised $1 billion or more. Pittsburgh-based Skild AI raised $1.4 billion for its robotics foundation model, Skild Brain.

Germany's headline deal came from Neura Robotics, which raised $1.4 billion for cognitive robots built with AI and data infrastructure.

Why is 'physical AI' different from chatbots?

The boom runs on more than compute. Ulrik Stig Hansen, co-CEO of San Francisco-based Encord, which provides the data layer for physical AI, laid out the bottleneck at Crunchbase News's inaugural Long Horizon Physical AI Summit in San Francisco earlier this month.

"We're just seeing physical AI take off," Hansen said. But building for the physical world takes time, especially compared with large language models that trained on the internet's existing text. "Physical AI is very different because there is no internet of physical AI data, and so the data first has to be collected, and you have to convert it from sensor tokens to action tokens," he said.

The data demands are punishing. Physical AI requires multimodal data sets — vision, tactile, force and torque — that are orders of magnitude larger than what text models need. "Storing all that data, curating all that data, enriching all that data, and validating that all the data works for these systems is a very expensive operation," Hansen said.

His timeline: the field is still in its pre-training phase, with a shift into post-training coming soon.

How fast is the unicorn class growing?

Roughly 75 robotics companies on the unicorn board have raised funding in the past three years. More than half of them — 45 — joined the board in 2026 alone.

The most highly valued startups cluster around two plays: the data and model layer, and vertical applications in heavy industry, manufacturing, logistics and household use cases, among others.

Hansen sees room for multiple winners. "There are a lot of companies that are raising capital, and I think a lot of them are going to win, just because there's so many different types of applications to the physical world," he said.

Funding is surging and new unicorns keep emerging, but the sector's direction remains uncertain as companies race toward further breakthroughs over the next year. The next test is whether that $9.9-billion quarterly pace holds once the data and model layer matures.