My opinion is that cool running can perpetuate a motor's lifespan indefinitely. This is a nominal 110VDC motor, the first mains-powered appliance motor ever marketed, whose original purpose for this example was likely to drive a sewing machine in a tailor shop.
Here it is run at very low power, about 10% of its rated voltage. 1887 original and unrestored,
I think the first ebike motor system was patented in the 1890s...and while this is not on-topic, the topic is life extension for our ebike motors of today.
Nothing lasts forever
(not a D cell, anyway, grin)
The lifespan of properly closely-fitted plain bearings, properly lubricated, run without dirt or water in the bearings, is nearly virtually forever.
The lifespan of ball bearings run clean with oil and not heavily loaded is also virtually forever.
The life of a motor winding depends on many variables, right? Yep.
There is thermal expansion stress, repeated countless times during its operating life, and that very thin layer of enamel or varnish that, if it fails at any one small spot where one turn crosses another, ends things.
Also, the higher the operating voltage design of a small motor, the finer is its magnet wire. And if hydrolysis, due to very high sustained operating temperature is acting as a slow, corrosive poison, the more certain it is that the wire will break somewhere and open circuit the motor.
I do not know if the hydrolysis mechanism (if it is even present) counts for much in the brief operating lifespan of an ebike motor. It may require thousands of hours to produce that effect. I have no experience or intuitive feel for hydrolysis failure and no actual case histories to refer to. Personally I have doubt hydrolysis, even if present, would harm your ebike motor in the bike's service lifetime.
Ten years ago, with my first ebike, a Currie with the outboard chain drive motor, I ran it overvolted, 36V instead of 24. Even with a healthy forced air ventilation of the little, brushed motor, it ran great and drove the bike at 25mph nearly, for....about 10 miles of a deliberate torture test. Then it suddenly fried.
The motor was about the same physical size as the little motor inside our Bafang cases, I suppose.
But the Currie failure is not indicative of what our Bafang hub motor can handle. Clearly, the Bafang, made in vast numbers, survives higher power outputs and passes away the resultant internal waste heat a lot better than that old Currie system, which was intended for only 15mph ebikes.
Does our Bafang and controller combination limit current if the motor temperature rises too high?
BTW, sidebar: that 1887 motor had accumulated many operating hours at full rated voltage, as evidenced by the deeply grooved (yet still very thick) commutator. The only "restoration" I did to the motor was to lathe level the commutator bar with a hand held blade, letting the motor run, cutting down the copper bars to get a flat working surface again. The laminated shim brass brushes were squared and trued. Why were the brushes made of such a relatively abrasive material then? Because graphite brushes were not yet invented. It was a few years more before graphite, still the brush standard today, was found to be the solution to the sole wear point of these early electric motors.