technology•Wed, September 16, 2026•5 min read

Google’s New Strategy: How Gemini Robotics Aims to Become the 'Android of Robots'

Google DeepMind is pivoting to a platform-first approach by embedding its multimodal Gemini AI into humanoid robots. By partnering with industry leaders, Google aims to standardize the 'intelligence layer' that powers the next generation of physical automation.

The Intelligence Layer

For years, the race to build advanced humanoid robots has been focused on hardware—the mechanics of walking, grasping, and maneuvering in physical space. But as robotic hardware matures, the industry has shifted its attention to the missing piece of the puzzle: the brain. Google DeepMind is now positioning itself to provide that brain through its new initiative, Gemini Robotics.

Rather than competing to build its own fleet of hardware, Google is pursuing a strategy reminiscent of its mobile phone dominance. By designing Gemini as a multimodal, general-purpose intelligence layer, the company intends to power robots built by a variety of manufacturers, effectively aiming to become the 'Android of the robotics world'.

Beyond Simple Automation

Recent deployments demonstrate that this vision is already moving into the real world. Google DeepMind has launched Gemini Robotics 2, a suite of models capable of controlling entire humanoid bodies—from legs to fingertips—enabling robots to perform tasks like crouching, reaching, and manipulating objects in unfamiliar environments. Key partnerships are already fueling this growth:

Boston Dynamics: Collaborating to integrate Gemini into the Atlas humanoid robot.
Apptronik: Utilizing Gemini Robotics to enhance the Apollo humanoid for logistics and industrial tasks.
Industrial Integrations: Working with companies like FANUC to allow human operators to instruct robots using natural language.

Google DeepMind’s robotics focus is building advanced AI that can power general-purpose robots. Automotive is a great place to start, but certainly our intention is to continue expanding to more and more applications.
— Carolina Parada, Senior Director of Robotics at Google DeepMind

Why the Industry is Paying Attention

The impact of this shift is significant. Traditional industrial robots were often rigid and required complex, specialized programming for every new task. The integration of Vision-Language-Action (VLA) models means these robots can now 'see' and 'understand' their environment in real time. This flexibility allows for rapid deployment in dynamic settings like factory floors, where conditions change constantly.

Furthermore, the data points show a clear trend toward automation. In 2024 alone, 542,000 industrial robots were installed globally, bringing the total operational stock to 4.7 million units. By providing an AI layer that can be updated and refined via software, Google is helping to turn these massive hardware investments into more versatile, long-term assets.

Key Takeaways

Google is positioning Gemini Robotics as a universal software layer for humanoid robots rather than building its own hardware.
Gemini Robotics 2 allows for whole-body control, enabling robots to navigate complex tasks like walking and manipulation.
Strategic partnerships with companies like Boston Dynamics and Apptronik are accelerating the real-world deployment of physical AI.
The transition to multimodal AI allows robots to move beyond hard-coded tasks toward more flexible, general-purpose intelligence.
The move follows a broader industry trend where cobots and humanoids are increasingly utilized to solve labor efficiency challenges.

FAQ

Is Google building its own robots?

No, Google DeepMind’s strategy is to create the 'intelligence layer' that other companies can use to power their robots, similar to how Android powers various smartphone brands.

What is the primary function of Gemini Robotics 2?

Gemini Robotics 2 acts as a vision-language-action (VLA) model that provides whole-body intelligence, allowing robots to interpret their surroundings and physically interact with objects more effectively.

Which companies is Google partnering with?

Google is collaborating with robotics firms such as Boston Dynamics and Apptronik, as well as industrial giants like FANUC.

How does Gemini Robotics differ from previous robotics AI?

Earlier robotics AI often required complex, task-specific programming. Gemini uses multimodal AI to allow for more natural instruction and better adaptation to dynamic, real-world environments.