From CDC and the Atlantic/Covid Tracking Project.
Figure 1: Weekly fatalities due to Covid-19 as reported to CDC for weeks ending on indicated dates (black), excess fatalities calculated as actual minus expected (teal), fatalities as tabulated by The Covid Tracking Project/Atlantic (dark red), all on log scale. Source: CDC accessed 8/14/2020 vintage, Covid Tracking Project/Atlantic accessed 8/15/2020 and author’s calculations.
Three observations: (1) the official CDC count for recent weeks is likely to rise toward the unofficial count as certificates arrive, (2) the unofficial count remains elevated at a high plateau of about 7470/week; (3) recent weeks’ (about a month’s worth) CDC data are subject to severe undercounting, so inferring recent trends on the basis of CDC data is not advisable; (4) the IHME forecast from 7 August were on the mark for the week just passed (according to the unofficial count). IHME is right now about in the middle of the pack of models surveyed by Fivethirtyeight. Current forecast is for over 200,000 fatalities by September 19th.
From week ending February 29th through the week ending August 1st (the latest CDC data, which incorporates severe undercounting in the last three weeks), the cumulative CDC fatality tally is 154.7 thousand. Cumulative excess fatalities is 204.7 thousand, implying an additional 50.0 thousand Covid-19 fatalities above the official tally over this period.
See also NY Times article, for a regional breakdown, and additional discussion.