1,000 humanoid robots. 1 factory.

<p>What does it actually take to put 1,000 humanoid robots to work?</p><p><br></p><p>Hexagon Robotics President Arnaud Robert joins me to explain how the company is moving its Aeon humanoid from pilots and demonstrations into real industrial production, including a plan with Schaeffler that could eventually scale to 1,000 robots.</p><p><br></p><p>And the big lesson is that scaling humanoids isn&#39;t about taking one robot doing one task and multiplying it by 1,000.</p><p><br></p><p>It&#39;s about building a multipurpose fleet that can move between jobs as demand changes across a factory.</p><p><br></p><p>We talk about:</p><p><br></p><p>• Why humanoids need to be multipurpose to justify themselves </p><p>• How Aeon gathers training data while doing real work </p><p>• Hexagon&#39;s “sim-to-real-to-sim” training loop </p><p>• Why customer-specific factory data matters </p><p>• The difference between a successful pilot and a production-ready robot </p><p>• The two metrics Hexagon cares about most: cycle time and human intervention </p><p>• What Hexagon learned from early sensor-fusion and actuator challenges </p><p>• Why Aeon uses wheels instead of legs — and can still climb stairs </p><p>• How humanoids could eliminate bottlenecks instead of simply replacing workers </p><p>• Schaeffler&#39;s “robot gym” and the train-deploy-scale path to 1,000 humanoids</p><p><br></p><p>Robert&#39;s view of the factory of the future isn&#39;t a lights-out facility with no people. It&#39;s a highly autonomous factory where humans and robots work together — with robots handling repetitive tasks and shifting between bottlenecks while skilled workers focus on problem-solving.</p><p><br></p><p>This is what humanoid robotics looks like when the conversation moves beyond prototypes and into production.</p><p><br></p><p>00:00 What does it take to deploy 1,000 humanoids?</p><p>01:31 Why 1,000 robots changes the problem</p><p>03:02 Why humanoids need to do multiple jobs</p><p>03:55 The race for physical AI training data</p><p>04:51 Simulation, synthetic and real-world data</p><p>06:20 Hexagon&#39;s sim-to-real-to-sim loop</p><p>08:10 What robot-generated training data captures</p><p>10:21 Pilot vs. production-ready humanoids</p><p>12:14 The industrial experience advantage</p><p>13:12 The two metrics that really matter</p><p>15:39 What Hexagon learned from failure</p><p>17:35 The humanoid actuator problem</p><p>18:47 Why Aeon uses wheels instead of legs</p><p>19:59 Yes, a wheeled human