I think I do need to clarify a few things here...
First of all, no matter what size of heat sink you use, the heat sink is NOT designed to "absorb" heat long term...
It is the job of the heat sink to absorb heat qui9ckly - then spread it around and somehow "get rid of it".
A heavier heat sink, which has more thermal mass, will take LONGER to heat up, and can absorb more heat over the short term...
But, in the end, no matter how much it weighs, a heat sink will continue to gradually get hotter and hotter over time, unless it can dispose of enough heat to reach thermal equilibrium.
In general heat is disposed of in one or more of three ways: conduction, convection, and radiation.
Conduction only works by direct contact...
However, unless it is infinitely large, whatever you're conducting heat to will eventually itself get hot.
And conduction also relies on what you're dumping heat into being a good conductor of heat.
For example, copper conducts heat quite well, titanium not so well, and wood doesn't conduct heat well at all.
Convection works on the principle of "hot stuff expands, gets lighter, and rises, which causes the heat to go up, and new cool stuff to move in from below".
Most normal heat sinks work by conduction and convection.
The heat sink gets warm, which warms the air touching it (by conduction), the warm air then rises (convection), is replaced by cooler air, and the process continues.
It is the job of the heat sink to conduct heat away from the devices generating it and provide surface area whereby that heat can be transferred to the surrounding air.
Because the forces that drive convection aren't very strong there is a balance between providing good contact with the air and avoiding interfering with the convective flow of air.
Another factor is that convection is driven by the difference in air temperature; therefore convection requires that something be allowed to get warm in order to make the air surrounding it warm enough to rise.
Convection is also a relatively weak force ; hot air rises but, because it isn't that much lighter than cool air, it doesn't try all that hard.
That's why we orient fins vertically instead of horizontally; because they allow the air to flow more freely so even that weak force can cause a lot of air to move past the fins.
Adding a fan to create a forced air flow bypasses this need because it forces air to travel past the heat sinks where heat can be transferred to it by conduction.
So... a properly designed heat sink should be heavy enough to absorb short bursts of heat into its thermal mass...
Then it must have good enough thermal conductivity to spread that heat over its entire area...
And it must have enough area to then conduct that heat into the surrounding air...
(Aided either by convection or air being forced past it by a fan.)
And, no, I didn't forget radiation.
The ability to radiate heat is directly proportional to temperature and is a relatively minor contributor at the temperatures inside solid state equipment.
(The vacuum inside vacuum tubes is an insulator as regards conduction and convection.
Tubes operate at much higher temperatures - and are cooled almost entirely by radiation.)
Thermal grease is an excellent conductor of heat.
And, in normal use, thermal grease does not really lose effectiveness, even when it dries out a bit.
Therefore, as long as it remains in place, you really don't ever need to worry about or replace it.
(The thermally conductive parts don't really dry out.)
However, when you apply thermal grease, the idea is for it to fill the minor gaps between the device and the heat sink, establishing perfect contact with both.
However, because not even thermal grease conducts heat as well as actual contact, you also want it to spread into as thin a layer as possible.
So, if you remove a component, and then replace it, there often simply isn't enough grease remaining to do the job...
And, if the grease has become stiff or lumpy, it may actually prevent the component from seating properly.
In other words, if you remove or replace a part, it is always a good idea to apply new grease.
However I don't know anyone who recommends removing parts simply to apply new grease "because the old grease is getting too old".
(Of course, once you remove that CPU or other component to "check" the grease, you have now created a situation where it should be replaced.)
Also note that there are other solutions and combinations...
For example those grey rubber "sil-pads" are a thermally conductive silicon rubber...
They entirely replace thermal grease in some situations and may be supplemented with it in others...
I should also point out that, while SOME products with long warranties are actually designed for a longer service life, that is NOT always the case.
Providing service for a product for the duration of its warranty is simply one of the manufacturing costs associated with producing and selling that product.
In many cases, it's perfectly reasonable to offer a longer warranty, even on a product that isn't more reliable (as long as you can offset the cost of doing so by charging a higher price).
(A warranty doesn't guarantee that a product won't fail... any more than health insurance guarantees that you won't get sick... instead both actually serve to offset risk.)
And there's the "sans fans" solution - build the cooling sinks with enough metal that they can absorb all the semiconductor heat. This is NOT a cheap solution, but it does work.
It also needs to be mentioned that no matter how the heat sinks are cooled, the semiconductors need to be able to TRANSFER their heat or else they don't do well. That heat transfer from semiconductor to heat sink is almost always accomplished with "heat transfer grease" that's applied during manufacture. All transfer grease has some volatile compounds in it. Over time, that grease dries out and its thermal conductivity drops significantly if not exponentially. The generally-accepted service life for heat transfer grease in computer semiconductor service is five to seven years. The semiconductors will still function, but they'll operate at significantly higher temperatures. Eventually, the higher temperatures (combined with the standard bathtub curve of reliability) will cause semiconductor failures.
Some heat transfer greases are better than others. What grease is used (and how quickly it dries out) depends on how much heat must be transferred and on the composition of the grease itself. For a computer CPU or GPU, the circuitry is dense, the heat generation per area is high, so much so that the heat sinks are normally supplemented by fans. The grease in those applications is stressed by the high temperatures and the large amount of heat per area that must be dispersed.
Audio output devices (unless run at consistently high output) normally operate at much lower temperatures, and the grease can remain stable for longer periods of time. The companies that most often offer decades of warranty avoid thermal failures by:
Using multiple output devices so that each transistor generates less heat
Using very high-quality thermal transfer grease to avoid dry out
Using a LOT of heat sinking material (or chassis cooling fans) to ensure lower operating temperatures
Including thermal monitoring circuitry that will either reduce power output or trip the amplifier off line to avoid thermal failures
These options are not inexpensive which is one of the reasons why power amplifiers with decades of warranty are usually more expensive than those amps with shorter warranty periods.
Boom
