Alphabet's robotics software development companyIntrinsicEarlier, it was officially announced that it will be integrated into the Google ecosystem. In the future, it will combine the Gemini AI model with Google DeepMind technology to create a "robot version of Android" that can be used across hardware platforms, and accelerate the application of "physical AI" in the manufacturing and logistics industries.
According toOfficial announcementIntrinsic, founded in 2021 and originally part of Alphabet's "Other Bets" (innovation experimentation) division, separated from Alphabet's X Labs in 2024 and, after collaborating with NVIDIA on related technologies, recently announced that it will be integrated into the Google ecosystem.
Integrating Gemini and DeepMind to completely reshape "physical AI"
Following its integration with Google, Intrinsic will continue to operate independently, but will receive greater technological support from its parent company. The company stated that future development will deeply integrate the Gemini foundational model with Google Cloud's cloud computing capabilities, and will collaborate closely with Google DeepMind, which is responsible for forward-looking AI research.
Hiroshi Lockheimer, Google's top product lead for "Other Bets," said in a statement: "We are thrilled to bring the Intrinsic team to Google, which will enable us to deliver groundbreaking AI technologies to more businesses and industries on a larger scale."
Intrinsic defines its core business as "physical AI"—a field that deeply integrates software and hardware, enabling machines to perform valuable tasks in the real world. In traditional manufacturing and logistics environments, robots can typically only execute single, hard-coded instructions; however, with the introduction of AI, robots will possess "perception" and "reasoning" capabilities, automatically adjusting their actions based on changes in part placement or production line conditions, thereby significantly improving factory productivity and flexibility.
Flowstate platform: Serving as the "Android system" for the robotics industry.
To achieve this goal, Intrinsic put forward an ambitious vision: to become "the Android of robots".
In the past, industrial robots from different brands (such as Yaskawa, Fanuc, and ABB) often had their own closed operating systems and programming languages. Intrinsic hopes to break down this barrier and establish a "common software infrastructure" that can span different brands of robot arms, cameras, sensors, and AI models.
To address this, Intrinsic launched a web-based development and simulation environment called "Flowstate." Developers don't need to write low-level code from scratch. Instead, they can quickly test robot behavior modules called "Skills" (such as component visual recognition, automatic grasping path generation, or torque sensor-based precision control) in a virtual environment and seamlessly deploy them to physical robots by dragging and dropping them together.
Analysis of viewpoints
The merger of Intrinsic with Google demonstrates that the competition among tech companies in the "Embodied AI" field has entered a fierce stage.
Over the past few years, while everyone was focused on pure "digital brains" like ChatGPT or Gemini, NVIDIA had already quietly laid a vast network in the fields of industrial manufacturing and simulation through its Omniverse platform and Isaac robotics system. Google's move to bring Intrinsic back to its own fold, and to brand it with DeepMind and Gemini technologies, is clearly aimed at directly competing with NVIDIA in the industrial automation market.
The phrase "creating the Android of robotics" highlights Google's strongest business model: I don't manufacture the hardware myself, but I provide the best free or open-source operating systems, making all hardware manufacturers dependent on my ecosystem. As smartphone growth peaks, the next wave of terminal devices capable of handling the computational demands of trillions of AI devices will undoubtedly be industrial robots in factories and future humanoid robots for home use. Whoever masters the "OS" (operating system) of these robots will reign supreme in the next decade's physical AI revolution.
