USA has had a 7 day average of around 40000 new cases a day for the month of september, and around an average of 750 deaths a day (more actually but for sake of argument.) That's a 2% CFR.
Intuitively, Id think the infection rate is underreported relative to the death rate.
More than likely but it's probably not an order of magnitude difference.
In the early summer, the CDC published a study estimating that the confirmed cases were 6-24 times lower than the actual cases. With testing continuing to ramp up, I suspect the gap has narrowed, but some significant multiple is still likely the case here and an order of magnitude is not outside the realm of possibility.
If 90% of cases are unsymptomatic or have minimal symptoms we'd expect an order of magnitude underreporting of cases, assuming people were only tested when they go to the doctor/hospital.
If 90% of cases are unsymptomatic and not showing up in the cases, then 75 million people in the US will have already had covid, and you're also getting 400 thousand new cases a day.
That's not how CFR works. CFR is fatalities from known cases. If you're thinking about infection rate, you need to look at IFR, not CFR. If you want to compare one disease to another, you need to compare CFR to CFR, and IFR to IFR.