I ask this every time I read a review of one of these monsters. No ones worried about hanging 1.5kg from a MB that's a couple of mm thick?
Answering more than one response here . . .
It's not so much the weight, but the torque. Without fans installed, the pipes and fins are insignificant. For the weight of the heatsink base, no torque there, so no problems.
Now if you add fans, there are ways to provide the same CFM throughput without putting the fan-weight on the cooler. First, duct the case rear output fan to the back of the cooler with a construction of foam art-board, or possibly -- by purchasing a silly-rubber accordion duct by ThermalRight.
I think your biggest risk with one of these coolers is the shipping or moving of the entire computer, so you might remove the fans before doing so.
I might have switched to AIO coolers almost ten years ago. I might have tried custom water-cooling. I was investigating "Rube-Goldberg" evaporative cooling systems. But when it came down to the decision, considering the maintenance and trouble in constructing, the occasional story of leaks and pump failure, I remained committed to air-cooling. At some point, we had CPUs with an IHS fitted to the processor die with a bad paste TIM, so the processors -- stock from the factory and instead of indium solder -- ran hot. Thus emerged an outfit in Texas named "Silicon Lottery". They performed various services. But if you had purchased the CPU in the intel box, they would de-lid it, replace the TIM with Thermal Grizzly Conductonaut after applying Liquid tape to appropriate areas, and ship it back to you for $50 -- with the result of as much as 20C reduction in load temperatures. Meanwhile, the CPU makers have returned to indium solder, or they never stopped using it. It was, as far as I know, exclusively an Intel problem.
So I see all the drawbacks of massive air-coolers in some of the posts here, but I've remained undaunted. A good AIO cooler, for instance, has a pump -- obviously. If it has a pump, then it has a motor, and if it has a motor, it has moving parts. And therefore, it has a limited and finite MTBF. The MTBF for a heatpipe cooler, according to NASA, is 1 million years.
If you don't want to bother with top-end air-cooling, the rumors and leaks about the forthcoming Arrow Lake desktop flagship CPUs include the idea that they will use 100W less power at full load than Raptor Lake -- and therefore 100W less thermal power.
If I can convince myself that I will never want to overclock my Arrow Lake, I can compromise my choice of coolers. Easily.