Adaptive Machines: The Founder’s Perspective

The trajectory of robotics has been shaped by rapid progress in computation, perception, and artificial intelligence. Robots today can plan paths, interpret environments, and adjust strategies in ways that were once limited to theory. Yet when these systems move from simulation into the physical world, the results often fall short.

The actuator bottleneck

The cause is consistent. Robots remain constrained by their actuators. The prevailing motor–gearbox paradigm was designed for steady motion. It performs well on factory lines but introduces inefficiencies and limitations in dynamic, variable, or human-centered settings. Motors consume power to generate torque even when no useful work is done. Gearboxes add inertia that slows response and reduces efficiency. These mechanisms cannot provide the passive, adaptive forces that natural systems rely on to achieve efficiency and resilience.

A pattern across research

Over years of research and collaboration in robotics laboratories across the United States, the United Kingdom, Germany, and Singapore, the same pattern emerged. Algorithms grew in sophistication, but hardware restricted what was possible. Breakthroughs in simulation and control often reached a dead end once physical prototypes were built. The bottleneck was not in the software but in the actuators that formed the foundation of every system.

The Adaptive Machines approach

Adaptive Machines was created to address this problem at its root. The company develops adaptive actuators with Physical Intelligence—hardware that makes use of mechanical adaptation rather than resisting it. These actuators can store and release energy in cyclic tasks, adjust stiffness for stability or compliance, and reduce the continuous power demand placed on motors. By aligning design with the underlying physics, they allow robots to operate more efficiently, more safely, and with greater capability.

Robotics at an inflection point

This work is motivated by a clear observation: robotics is reaching an inflection point. Collaborative robots and humanoids are moving from prototypes to early deployments. Artificial intelligence has advanced to the point where decision-making is no longer the limiting factor. Yet scaling these systems requires hardware that is as capable as the software controlling it. Proprietary, custom-built actuators can serve individual demonstrations, but they cannot support widespread adoption. Standardized adaptive actuators can.

The mission ahead

At Adaptive Machines, the focus is singular: remove the actuator bottleneck and provide the hardware foundation that robotics requires. By rebuilding from the ground up, starting with the actuator, we create a path toward robots that are not only intelligent in computation but also intelligent in their mechanics.

— David Braun, Founder of Adaptive Machines