Robotics gurus and engineers: what is the disadvantage of shifting weight to the torso in a hexapod robot?

I’m thinking about designing a hexapod robot, mostly as a mental exercise, but perhaps one day as an actual build.

I’d like to build one large enough to handle stairs and so it will be significantly larger than the toy versions available on the market. I’ve notice that on the toy versions there is a servo at each the 3 joints of the hexapod leg. This makes perfect sense for simplicity sake.

This does add substantial weight to the leg segments (especially with larger motors and 3d printed cycloidal drives), which puts extra strain on the femur axis since it has to raise the weight of the femur, tibia and the tibia’s actuator. My idea was to move the servo for the tibia joint next to the femur and use a timing belt to actuate the tibia. Essentially all 3 motors would be stacked around each other in a cluster near the coxa joint. I would be taking inspiration from this quadruped’s leg design. This means much less torque is required to actuate an individual leg, but also that the torso becomes significantly heavier.

My thinking is that this additional weight at the torso will be distributed across 3 other legs (at minimum), since all hexapod gaits require at least 3 feet contacting the ground at any given time. I came across an old thread somewhere on the internet where shifting the weight from the legs to the torso in the design was considered a negative of the design by several people without much additional discussion. That made me realize that I’m probably not understanding something important about the weight distribution. Would an ideal hexapod robot design not have light legs and a heavier torso, all things being equal? What am I missing/not understanding?