top of page

Extend Robotics joins the Google DeepMind Accelerator: building the next era of Robot‑as‑a‑Service

Today we're excited to share a major milestone: Extend Robotics has been selected to join the Google DeepMind Accelerator program.


For our team, this isn't just a badge, it's an opportunity to deepen what we've been building from day one: practical, scalable robotics that creates measurable value in the real world. And we believe the fastest path to that future is Robot‑as‑a‑Service (RaaS) - robots delivered as outcomes, not as one‑off hardware deployments.


About the Google DeepMind Accelerator


The Google DeepMind Accelerator is a three-month, equity-free program for early-stage robotics start-ups across Europe. Each cohort brings together 15 companies for 12–15 intensive weeks combining digital training, hands-on workshops, and 1:1 mentoring from Google AI and robotics experts plus access to the full Google AI stack, Gemini Robotics Models, and up to $350k in Google Cloud credits.


Our cohort kicked off with a 4 day in-person event in London in June 2026. Graduation is a 3-day event in London in September 2026, culminating in Demo Day, a showcase to Google's network of mentors, partners, and teams.




Why RaaS and why now?


Robotics is at an inflection point. Hardware continues to improve, but the real unlock is happening in software: better perception, more robust planning, faster learning cycles, and tighter feedback loops between deployments and model improvement.


At the same time, more companies want robotics, yet many are still blocked by integration complexity, operational overhead, and ROI uncertainty. RaaS changes that equation: lower-risk adoption, faster time-to-value, and performance measured in outcomes. Success in RaaS is less about the "robot moment" and more about the repeatable machine behind it - scoping, deployment cadence, feedback loops, and operational structure that compounds over time.


What we believe about where robotics is headed


The accelerator sharpened our thinking. Here are the convictions we're carrying forward.


We do the whole chain - from teleoperation to autonomy.

Extend Robotics is one of the few companies doing the complete chain: from teleoperation through to autonomy while leveraging existing robot platforms rather than custom-building everything. Many competitors operate only at the pipeline layer. We do the whole thing. That breadth isn't accidental; it's what lets us deliver outcomes rather than components, and it's what scales.


Pilot purgatory is a design failure, not bad luck.

Too many robotics deployments stall between proof-of-concept and production. The cause is almost always the same: the pilot lacked a defined strategic outcome and clear success metrics. Every deployment we run is anchored to a specific, measurable result. That discipline is what moves a customer from a promising demo to a scaled fleet.


We design for the factory floor, not the lab.

Most industrial facilities run 24/7 and cannot take downtime. Robotics today has to integrate into existing operations rather than replace them—and that integration cannot stop the line. This is why task selection is so critical. Success means deep respect for the existing ecosystem, honest scoping, and not overpromising what AI will solve on day one.


The roadmap starts with what's actually possible.

The core problem in robotics today is that customers don't yet understand what it can and cannot do, so they over-expect, and disappointment follows. We're building a defined set of repeatable use cases drawn from our strongest deployments. This lets us be genuinely honest about what isn't possible yet, sell a credible roadmap for what is, and maintain the kind of trust that turns pilots into long-term contracts.


The use cases are also designed to extend: picking a box from A to B and picking a bottle from A to B are essentially the same problem, as long as we're not adding heavy classification or detection complexity. Generality at the task level is what makes RaaS unit economics work.


The Google ecosystem matters beyond compute.

Access to Google's AI stack, including Gemini Robotics Models, is genuinely valuable. But the deeper advantage is the network: the right technical conversations, the right doors, the right infrastructure partnerships to build processes that scale. This is why the accelerator matters to us strategically, not just technically.


What's next


We'll share more soon on:

  • The RaaS use cases we're prioritising and how we scope them

  • How we define and measure outcomes in production deployments

  • What it takes to move from a first pilot to a scaled fleet operation

  • Learnings from working alongside the Google DeepMind ecosystem through to graduation


Extend Robotics is here to collaborate and to ship real robots doing real work.




bottom of page