The Deployment Moat: A Physical AI Industry Panel

As robotics foundation models and hardware keep advancing, real-world deployment remains one of the industry's biggest challenges. Our third flagship asked where the moat really sits: in the technology, the deployment expertise, the operational data, or the customer relationships built along the way.

Robert Scoble opened with a fireside on what is actually running versus what is still a demo. Two panels followed: who owns the moat, with an investor, a robotics platform and a robots-as-a-service operator; and how to break the deployment bottleneck, with teams building robot intelligence, deployment infrastructure and dexterous hardware.

What we covered

• 01Opening fireside: what's actually running, and what's still a demo
• 02Panel 1 · Who owns the moat? Deployment as competitive advantage
• 03Panel 2 · From demo to floor: breaking the deployment bottleneck

Eight takeaways

1. 01Reliability is a product of every layer, not a property of the model
2. 02The customer is buying an outcome, so someone has to own the risk
3. 03Agree on what success means before the robot arrives
4. 04Your first customers are the ones who need it most and can decide fastest
5. 05Measure usage, not revenue
6. 06Plan for humans in the loop, permanently
7. 07In the field, every extra moving part is a liability
8. 08Narrow, repeatable sites are what make deployment compound