Google DeepMind gives 15 robotics startups Gemini Robotics access
Original: Google DeepMind opens Gemini Robotics support to 15 European startups View original →
Fifteen startups get closer to the model stack
Google DeepMind is turning its robotics work into a structured channel for European startups. In a post published at 2026-06-12 11:01:22 UTC, the account said “15 startups” had joined its Robotics Accelerator and would receive access to its AI stack, Gemini Robotics models, and support from Google teams. The number matters because this is not just a general call for applications. It is a defined cohort that can test foundation-model robotics against messy real-world constraints.
“Our Robotics Accelerator has launched with 15 startups.”
The Google DeepMind account is one of Google’s primary channels for Gemini, research, and robotics updates, so the post is worth treating as first-party signal. The linked Google blog adds the useful context: the three-month program starts with founders gathering in London and spans companies from Norway, Greece, Romania, the United Kingdom, France, Germany, Switzerland, Italy, Denmark, and Sweden. The selected companies cover robotic welding and metal 3D printing, industrial manipulation, device testing, construction MicroFactories, ocean robots, waste sorting, humanoids, teleoperation pipelines, machine monitoring, high-resolution robotic touch, and even microrobots for brain tissue.
Why access beats a demo clip
The technical stake is the connection between Gemini Robotics and deployment. Robot companies do not only need a stronger model; they need data loops, safety checks, manipulation policies, simulation, product guidance, and a path from lab behavior to repeatable customer work. Google says the startups will receive technical mentorship, product guidance, and partner access, which makes the accelerator a useful probe of where embodied AI is ready for production and where it still breaks down.
What to watch next is whether any of the 15 teams report measurable deployment gains after the three-month program: faster setup, better manipulation reliability, fewer human interventions, or cheaper data collection. Google DeepMind will also be testing its own market position. If Gemini Robotics becomes a default layer for young robotics companies in Europe, the program could matter more than a single model demo. Source: Google DeepMind source tweet · Google blog
Related Articles
Google DeepMind has opened applications for a Europe-focused robotics accelerator. The three-month program combines technical mentorship, product support, and potential cloud credits up to $350k for eligible startups.
Google DeepMind and Agile Robots said on March 24, 2026 that they will combine Gemini Robotics foundation models with Agile Robots hardware to build adaptable, reasoning robots for industrial environments. The partnership starts with high-value manufacturing and automation use cases where scale and reliability matter most.
NVIDIA released open-source physical AI agent skills across Omniverse, Cosmos, Isaac, Metropolis, Alpamayo, and Jetson. The company points to manufacturing gains including 67% faster training and deployment at Pegatron and a 17% detection-rate improvement at Delta Electronics.