As someone who has worked with both Unitree and Boston Dynamics I can tell you there is a marked difference in the ability for Boston dynamic's ability to walk vs Unitree's. The ANYMal is closer to Boston Dynamic's capabilities but also at a similar price point. That being said what Unitree is lacking is good software so theres a chance that they may be able to incrementally catch up while maintaining a relatively low pricepoint. I don't think we're gonna see a DJI moment any time soon in this space purely because the other algorithms around this area are still in their infancy. Furthermore, motor prices don't seem to have fallen all that much making innovation in this space much harder than with drones.

Forgive me if I'm wrong but I though motors aren't the main cost, rather the rest of the actuator?

Also Unitree is going to make so much money selling these as well as get so much mass producing experience they'll probably be able to figure out the algorithms. Methinks the code is easier to figure out than reliable and cheap hardware.

The motor usually makes up about a third, the motor controller another third and the planetary gears and enclosure the final third.

You can see this in the Unitree videos, nearly everything is on a flat plane and while it was a video from 2019 it seemed to struggle climbing stairs a bit.

But they'll sell at a higher volume and be able to reinvest in R&D. Boston Dynamics needed constant injection of capital.