No way we're only at 20% infection with over 15.5M identified cases. That would be about 66M cases, we've gotta be beyond 100M infections at this point. Less than 7x identified cases, rational multiplier.
I do think the unreported cases are many times reported cases. I don't think we're as high as 100 M infection yet though.
The problem with all these things is there are many variables and because of all the screw ups in the US, most measurements have a lot of uncertainty, especially case numbers. In countries with good record keeping, the deaths can be accurately tracked, but even in some of those countries some numbers might be under reported for political reasons (such as some parts of the US). Some countries have more vulnerable populations than others due to age of the population, life habits in the population, and the number of risk factors in the population (such as levels of Type II diabetes and obesity).
Some countries known for good testing and good health care systems have brought the virus under control such as South Korea, Taiwan, and New Zealand, but the bulk of their deaths happened early in the pandemic when treatment regimes were not as good as today. If they knew then what we know now, some percentage of their deaths never would have happened.
The CFR for the three countries mentioned hover around 1%. Even if that's 2X what it would be today, that puts them around 0.5%. The US CFR is 1.9% with probably some under reporting from some states for political reasons. Some US deaths such as most in New York and New Jersey also happened early when treatment techniques were evolving, but more are happening today than back then. (I got my numbers from John Hopkins' reporting site.)
If the real CFR for the US is really 0.5% with up to date treatment techniques, then that would be an under reporting rate of about 4X max. Maybe 5X if the under-reporting of deaths is significant from places like Florida and Texas, but that's really stretching it. Considering that the US has a higher rate of comorbidities than the three countries above, the real CFR probably is higher, which would push the under reporting rate lower.
So I would say the US is probably in the 40-50 million total cases range with a low probability it might be as high as 70 million. That works out to about 12-15% herd immunity at this point. It could be lower than 40 million due to the factors cited above. We don't know and probably won't know for several years until deep dive studies can be done by academics with hindsight and a much better data set.