This is an attempt at marketing that fails due to oversimplifying an engineering issue until it actually means the opposite of what it's supposed to mean. And this question is all over the internet...
As explained by the others, capacitor lifetime depends on temperature. I picked a random datasheet (Nichicon R5), this is the "endurance" spec:
Test condition 105°C, rated voltage 2000Hrs.
Capacitance change Within ±20% of initial value before test
tan δ 150% or less than the initial specified value
ESR 150% or less than the initial specified value
Leakage current Less than or equal to the initial specified value
This means the manufacturer guarantees that after a number of hours at a specific temperature, the important specs (capacitance and ESR) will not degrade by more than -20% and +50% respectively.
It doesn't mean the cap pops off or stops working completely, it simply degrades. If the application does not require the full performance of the capacitor, then it may work fine for a much longer time. For PC mobos this is not the case, high performance caps are required on CPU VRM, especially ultra low ESR due to the huge ripple currents involved. ESR is equivalent series resistance, and it will increase as the capacitor gets old or if the electrolyte dries out, which makes the capacitor heat more (due to resistive losses) and perform worse at its supply voltage smoothing role.
This is different from the well known "capacitor plague" which was due to a faulty electrolyte formula. In this case the capacitors did not honor their lifetime specifications, they were defective products which are no longer manufactured.
Now, Alu-polymer caps rated for 12k hours at 105°C are very good quality components, some of the best available. Usually you would get 2k hours 105°C caps, so Asrock is selling a good product with dumb marketing.
However it doesn't matter at all. Using the lifetime estimation from Illinois Capacitors that Tim posted, at 60°C (which is already very hot for a mobo, time to clean the dust!) the polymer 2000h/105°C cap would last 40 years already, and the 12000h cap would last two centuries. Temperature has a huge influence!
So, yeah the 12k hour caps are a nice touch but they're a bit of a luxury. 2k hour 105°C caps would still work fine when the computer becomes obsolete and is replaced. Especially since the 60°C temp I used is quite pessimistic, the caps should be much cooler than this unless you run high cpu load 24/7 and the airflow is terrible. So, no planned obsolescence here.
The caps that are accused of "planned obsolescence" and die are usually the ones which are inside consumer electronics power supplies, right next to a very hot heat sink with low airflow, they're run close to max spec and aren't polymer caps but rather low-Z alu electrolytics which are more vulnerable to heat. If the manufacturer puts a crapxon brand 85°C-rated cap in there, yeah it's gonna go "planned obsolescence" pretty quick. Also the caps on the HDMI board in a buddy's AV receiver, all 85°C caps next to a big hot CPU with no fan and no airflow, this receiver is famous for this, all the caps always die. This is more of a design mistake or excessive cost-cut than planned obsolescence though.
7Doesn't "20% longer" sound a lot better than "not quite as crappy as some others"? – fixer1234 – 2018-02-16T23:26:27.953
3Agreed, but I have boxes with high-end boards I built 10 years ago that are still fit for daily use (back when Abit was still in the picture). I'm just having trouble mapping this statement to reality. – Frank Thomas – 2018-02-16T23:29:51.373
3@FrankThomas: "lifespan of at least 12,000 hours" doesn't mean it gets to 12,000 hours then stops. It means if it stops before 12,000 hours then it is treated as a quality control problem. My TV's manufacturer's warranty lasted I think 12 months? I've had it for 15 years. Now of course lifespan and warranty aren't the same thing, but the principle is roughly the same.. – Lightness Races with Monica – 2018-02-17T01:37:47.263
1
Might be a question for Electrical Engineering Stack Exchange. Also see the Capacitor Plague, where Japanese capacitors were a selling point. It was a big deal back in the 1990s.
– jww – 2018-02-17T05:33:28.1731.37 years MTTF would be a terribly unreliable mainboard. For comparison: Server mainboards more often than not will make 10 years intact - though they use somewhat more strictly selected components and are subject to better cooling discipline (see what is said about temperature...). – rackandboneman – 2018-02-18T21:59:58.070