Capacity would be aggregate bandwidth, which is determined by the satellite's communications technology (number of transponders, spectrum allocated, modulation used, etc) rather than directly by which orbit it is in.
Comparing constellation sizes does make sense in determining total capacity (aggregate bandwidth of the constellation) assuming that Starlink and Kuiper use similar capacity satellites. I think that is a reasonable assumption, but we have limited info on what Kuiper are planning. Current Starlink satellites have on average 20 GBps capacity each. I can't find any public info on the planned capacity of Kuiper's satellites. Even if Kuiper has higher bandwidth satellites, they'd need a fourfold advantage in bandwidth to overcome Starlink's fourfold advantage in satellite count. And SpaceX is likely to increase the capacity of Starlink in future iterations, so even if Kuiper did have such a bandwidth advantage, it might not last.
The satellite capability is one thing, but spectrum is another. SpaceX doesn't have priority on a lot of the spectrum they have, so it might be completely unusable by the time they go live.
I feel like we are arguing two different things. I was saying Kuiper can't beat Starlink. You are arguing Starlink is not going to succeed. I'm more optimistic about its prospects than you are, but even if you are right, it wouldn't change my original point – if both fail, then Kuiper still doesn't beat Starlink.
Other than Kuiper, Amazon has shown no signs of becoming a serious player in the consumer ISP space. Sidewalk is not a consumer ISP service in its own right, it needs to piggyback on a third-party consumer ISP service. Amazon is nowhere near becoming anything resembling Bell.