Alphabet has moved Intrinsic from its moonshot portfolio into Google, signalling that industrial robotics is now a core physical AI priority.
Intrinsic remains a distinct group but gains closer access to Google DeepMind, Gemini and Google Cloud. Its mission is to make industrial robots easier to program for manufacturers that do not employ large teams of robotics specialists.
The move combines models, development software and cloud infrastructure inside one organisation.
Flowstate aims to become an operating layer
Industrial robot hardware has become more accessible, but programming a new application can still require hundreds of engineering hours and code specific to one machine.
Intrinsic’s web-based Flowstate platform lets users build applications without writing thousands of lines manually. It is designed to remain independent of a particular hardware vendor, software stack or AI model.
CEO Wendy Tan White says Google chief Sundar Pichai compared the concept to Android: an operating layer that can support a wider ecosystem rather than one closed robot.
Intrinsic emerged from Alphabet’s X division in 2021 after five years of development. Joining Google now moves it from a separate bet toward infrastructure aligned with core AI strategy.
Google assembles a full physical AI stack
The company has recruited robotics leadership, released a standalone software-development kit and partnered with Boston Dynamics to bring Gemini into Atlas humanoids for manufacturing.
DeepMind supplies reasoning, perception and learning. Intrinsic provides industrial deployment software. Google Cloud offers training and operating infrastructure. Together, the pieces could help machines interpret sensors, adapt to changing environments and execute more complex work.
Intrinsic also acquired the commercial arm behind the Robot Operating System in 2022, connecting it to an important open robotics ecosystem. A 2025 partnership with Foxconn targets general-purpose intelligent robots for electronics manufacturing.
The source cites a McKinsey projection that general-purpose robotics could reach a $370 billion market by 2040. The forecast is distant, but it explains why major platforms want a position before standards consolidate.
Accessibility determines enterprise adoption
Manufacturers care less about a robot’s research pedigree than the cost and time required to deploy it. A platform that reduces specialist coding could make automation viable for more factories and more variable tasks.
Hardware-agnostic design can also reduce lock-in, although Google’s deeper integration may test how neutral Flowstate remains in practice. Customers will watch whether it continues supporting competing models and clouds as Gemini becomes more central.
Production economics remain the real benchmark. A simplified development environment must still deliver reliable motion, safety, integration and maintainability on a factory floor.
Bringing Intrinsic into Google gives the project resources and a clearer route to models and customers. It also raises expectations that the platform will move beyond experimentation into repeatable industrial results.
The significance is not an internal reporting change. Google is treating physical AI as a strategic layer where software can do for robotics what common operating platforms did for computers and phones.
If Flowstate lowers programming barriers while preserving hardware choice, manufacturers could automate work that is currently too specialised or expensive. Google’s bet is that the next major AI interface will not only answer on a screen—it will coordinate machines acting in the physical world.
That ambition makes safety and interoperability central. A software layer must represent different robot capabilities accurately, prevent an application from issuing unsafe motion and preserve performance when factory connectivity is limited. Manufacturers will also expect long support cycles that differ from consumer software. Intrinsic’s success depends on combining Google’s pace of model improvement with industrial expectations for stability. The Android analogy is attractive because it suggests scale, but factories will judge the platform by predictable integration, recoverable failures and whether one application can genuinely move between hardware without extensive re-engineering.
