Deaths don’t track 1:1 with hospitalizations because some people die much sooner. The most at risk tend to die within 2 weeks, while others are still extremely sick and can die 6+ weeks after infection. Those people have lower rates of death per day, but may still require ventilators. Thus, a steady rate of new infections can look like a very slow decrease in deaths even as the heath system is getting more stressed.
The number of tests per 1000 people is increasing each day in the US.
https://ourworldindata.org/grapher/full-list-daily-covid-19-...
This means that it is increasingly likely to uncover more cases, independent of how many new infections there actually are.
Where is the data on overloaded hospitals?
https://ourworldindata.org/grapher/new-covid-deaths-per-mill...
Measuring the number of cases as the testing is being massively bumped out does not measure overload - actual death rate does.