About 97% do. But it's kind of hard on the 2.1 million who will die achieving Paul's herd immunity level.
You would assume that 100% of the people would end up getting infected—and reality is that virtually no individual virus strain ever achieves that kind of infection rate.
Plus, we don't really have an accurate picture of how many people were infected in New York City because the antibody tests we use diminish rapidly in accuracy after a few months beyond the infection and we had virtually no PCR testing when the infection was spreading at its fastest. We do know it was circulating at least several weeks before any cases were confirmed and, at minimum, millions of New Yorkers had caught the virus without being officially counted.
You look at places that had their main surges later, when more widespread testing was in place, and you see deaths per confirmed case much closer to 1-2%, with of course the caveat that the risk varies widely by age and other co-morbidities.