Intrinsic joins Google to supercharge physical AI

Silicon Valley loves spinning out moonshots into standalone companies—until the AI race demands everyone pull in the same direction.

Alphabet-owned robotics platform Intrinsic just announced it is joining Google as a distinct operating group. Instead of operating as an outside Other Bet, the robotics software team will now plug directly into Google DeepMind, Gemini frontier models, and Google Cloud infrastructure.

Alphabet declined to disclose any financial terms or transaction details.

From moonshot spinout to the mothership

Intrinsic spent five years gestating inside Alphabet's X lab before graduating as an independent company in 2021, putting it in the same class as Waymo and Wing. Led by CEO Wendy Tan White, the team wasted little time expanding its footprint.

In 2022, Intrinsic snapped up Vicarious, a robotics startup backed by Jeff Bezos with roughly $250 million in venture funding. Months later, it acquired the commercial divisions of Open Robotics, the group behind widely used open-source robotics tools.

Even with a 20% workforce cut in early 2023, Intrinsic kept shipping. It rolled out Flowstate, a web-based platform that lets developers build robot workflows with modular skills rather than tedious low-level code. Late last year, it released its Intrinsic Vision AI model, followed by a joint venture with Foxconn to pursue full factory automation for electronics manufacturing.

Now, all of that industrial plumbing is coming under Google's roof.

Building the Android for factory floors

Industrial robots are notoriously rigid. Reprogramming an arm to lift an EV battery or inspect server trays usually eats up hundreds of hours of manual engineering.

Intrinsic's Flowstate aims to act like Android for physical hardware: an abstraction layer that works across different arms, grippers, cameras, and sensors. Adding DeepMind's research and Gemini's multimodal reasoning means robots can adapt to messy real-world variations on an assembly line instead of halting the entire line when a part is slightly off.

Why it matters

We think this folding back into Google is a telling reality check for the robotics ecosystem: standalone robotics platforms simply cannot afford the compute, talent, and foundation model infrastructure required for physical AI on their own.

For industrial developers and factory operators, this move makes Intrinsic significantly more viable long-term. Building factory automation on a platform tied directly to Google Cloud and Gemini removes massive integration headaches, making physical AI accessible without needing an in-house PhD team.

For Big Tech, physical AI is the new battleground. Leaders like Nvidia's Jensen Huang have made it clear that taking models off the screen and into the physical world is where the real enterprise value lies. Google already had frontier intelligence with DeepMind, but Intrinsic—and its major partnership with Foxconn—gives Google a direct pipeline into high-volume manufacturing floors.