It's very inaccurate. My estimated Apple VO2Max (based on heart rate alone) is 10 points less that my estimate from Garmin (based on a power meter and heart rate calculation).
How do you know the Garmin isn’t the inaccurate one? Because it used a power meter? Unless a tube is stuck in your mouth for you to breath through, they are all just “predicted VO2Max”, not measured. I don’t know which is more accurate, but I would hesitate to favor one over another without baseline against a real VO2Max test. However, like a weight scale, as long as it is accurate versus itself it is still a useful number.
Even for running without a power meter a Garmin device can give you a fairly accurate VO2max estimate if you do an occasional max effort time trial. But it has to be on a flat course: the FirstBeat Analytics algorithm doesn't account for hills. I have had a real VO2max test with a breathing mask on a treadmill and the results were very close.
I'm aware how VO2Max will be calculated in a lab. But a powermeter is a closer approximation since the effort is captured. In cycling, if you don't have a power meter the estimate would be based on distance and climb and won't factor wind or other external factors.
Also, I know it's definitely greater than the Watch estimate. With a below average cardio I can't bike the distance or climb I do on average. I did 3 metric centuries last year (with climbs over 4000ft each), for example.
EDIT: though you might not be wrong: https://sites.udel.edu/coe-engex/2019/03/16/how-accurate-is-.... Summary: Garmin's algorithm from Firstbeat is pretty darned accurate.