Google Absorbs Intrinsic From 'Other Bets' — The Physical AI Play That Signals Alphabet's Factory Floor Ambitions

Google absorbed Intrinsic back into the core Alphabet structure on February 25, 2026, pulling it out of the Other Bets category where it had operated since its founding. This is not a minor organizational adjustment. Other Bets is where Alphabet parks speculative projects that are years from commercial viability: Waymo, Verily, Wing. Moving Intrinsic into Google proper signals that Alphabet believes industrial AI robotics has crossed from research into commercial readiness.

What Intrinsic Actually Built

Intrinsic developed the Flowstate platform, a software layer that allows industrial robots from different manufacturers to be programmed, coordinated, and managed through a unified interface. Traditional industrial robotics requires custom programming for each robot model, each task, and each production line configuration. Flowstate abstracts that complexity, allowing manufacturing engineers to define workflows at a higher level and have the platform translate those instructions into robot-specific commands. Think of it as the operating system layer for mixed-vendor factory automation.

The Gemini Robotics-ER 1.5 Model

The immediate benefit of the absorption: Intrinsic now has direct access to Google's Gemini foundation models. The company announced Gemini Robotics-ER 1.5, a version of Gemini fine-tuned specifically for industrial environments. ER stands for Environment and Robotics. The model processes visual input from factory cameras, sensor data from IoT devices, and natural language instructions from operators to generate robot control sequences. A manufacturing engineer can describe a task in plain language, and Gemini Robotics-ER translates that into coordinated multi-robot execution plans.

The Foxconn Joint Venture Changes the Scale

Intrinsic and Foxconn announced a joint venture to deploy AI-powered robotics across Foxconn's manufacturing facilities. Foxconn operates the largest contract electronics manufacturing operation in the world, producing devices for Apple, Amazon, Google, and virtually every major consumer electronics brand. The joint venture gives Intrinsic access to production environments running at scales that no other AI robotics company can match. If the Flowstate platform works on a Foxconn production line, it works anywhere.

Why This Matters for Industrial IoT

The absorption of Intrinsic connects Google's cloud AI infrastructure directly to the factory floor. Every robot running Flowstate generates data that flows into Google Cloud. Every Gemini Robotics-ER inference query creates compute demand on Google's infrastructure. Every manufacturing facility that adopts the platform becomes a Google Cloud customer for industrial workloads. This is the same playbook Google used with Android: give away the platform layer to capture the infrastructure revenue underneath.

The Competitive Landscape Sharpens

Amazon has been deploying robotics aggressively in its own fulfillment centers, with over 750,000 robots operational as of 2025. Microsoft invested in robotics simulation through its partnership with NVIDIA on digital twin platforms. But neither Amazon nor Microsoft has made the move Google just made: absorbing a dedicated industrial robotics company into the core business with direct access to foundation model capabilities. Google is betting that the combination of Gemini's multimodal AI, Intrinsic's robotics platform, and Foxconn's manufacturing scale creates a defensible position in physical AI that pure-software competitors cannot replicate.

What Manufacturing Leaders Should Evaluate

Three questions for any manufacturer considering AI robotics. First, does your current automation stack support multi-vendor robot coordination, or are you locked into single-vendor programming environments? Flowstate's value proposition is vendor-agnostic orchestration. Second, what is your data infrastructure for factory floor AI? Gemini Robotics-ER requires sensor data, camera feeds, and production system integration. The AI is only as good as the data pipeline feeding it. Third, what is your timeline for labor force transitions? AI robotics does not eliminate factory workers overnight. It changes the skill profile from manual operation to system supervision and exception handling. The workforce planning needs to start before the technology deploys.

What Could Go Wrong

Google's history with hardware-adjacent businesses is inconsistent. Google Glass, Google Fiber, and Nest all demonstrated strong technology that struggled with go-to-market execution in physical industries. Industrial manufacturing has long procurement cycles, high switching costs, and deep resistance to platform dependency. If Foxconn's joint venture succeeds, it validates the approach. If it stumbles on integration complexity or production disruptions, it could set back enterprise confidence in AI-powered factory automation by years.