I heard it word of mouth from people that were there, operating or maintaining the tanks, and it was widely reported.
The GAO report you linked is conspicuously missing failure rates, and doesn't contradict anything I said above. In fact, it supports my claims.
Turning on 90% of the time immediately after being repaired doesn't mean its running at the end of the day, or even an hour later!
From the GAO report you linked, they ran out of most types (60%) of spare parts and started cannibalizing tanks after only 4 days of operations. They had major computer issues ("communication issues" according to the report), and had to fall back on old fashioned radios and obsolete targeting.
In addition to the availability issues, these problems caused the US to accidentally or intentionally destroy 1% of its own tanks during the 4 day operation (the breakdown from the report is below).
The report also talks about how the recovery vehicles were unreliable and had major problems reaching broken down tanks. It makes it clear parts shortages were due to the tanks burning through a large stockpile of parts that had been shipped there before the operation began, not due to poor planning.
There's no way those tanks would have survived an operation that dragged on for months like Ukraine has.
However, the operation was a success despite this.
I think this supports my conclusion that superior targeting and logistics won that war, not reliable, robust hardware platforms.
The rest is pasted from the GAO report (I didn't correct their OCR errors):
Abrams crews indicated that its range was limited because it frequently had to stop to (1) refuel to compensate for high fuel consumption and faulty fuel pumps and (2) clean air filters due to extremely sandy conditions.
Bradley and Abrams crews reported problems obtaining repair parts, and many had exhausted their limited supply of some parts by the end of the loo-hour ground war. Because of these problems, according to some Army logistics personnel, sustainability could have become a major problem had the war lasted longer. Crews also experienced problems in posit,ively identifying enemy targets and in having to use outdated and unreliable radios.
...
For example, logistics personnel from the 1st Cavalry Division told us that about 60 percent of the parts they were authorized had zero balances by the end of the war. To compensate for the inability of the established system to provide needed parts, combat units had to search logistics bases for needed parts, to trade with other combat units, or to take parts from other vehicles. According to some Army personnel, the inability to replenish parts reserves could have impeded sustained combat operations in a longer war.
...
23 Abrams were destroyed or damaged in the Persian Gulf area. Of the nine Abrams destroyed, seven were due to friendly fire, and two were intentionally destroyed to prevent capture after they became disabled. Similarly, of the 28 Bradleys destroyed or damaged, 20 were due to friendly fire. Moreover, weapon system capabilities were not optimized because the weapons’ ranges were greater than the sights’ ranges. Crews also noted problems with ineffective radios and suggested that a navigation system be installed in every Bradley and Abram% Army officials recognized the need for improvements in these areas (see app. III).