so the short interest from August 14th has been published and the trend -- of gradual short covering -- reversed. The previous peak was April 15th which was followed by seven consecutive reporting periods of lowered short interest. Looking at the chart the high of $1643 was hit on July 20th and the price declined. By the prior report (July 31st) the stock price had leveled off in the mid 1400s, but then on August 7th the price started to decline, bottoming out on August 11th before climbing again -- either despite the increased short interest or it had reached even higher levels.
This is an unprecedented level of value at risk. Since the apparent $15B ceiling was disproven July 15th when it was nearly $20B this is now just a bit more than $20B. Of course, on the 14th the stock price closed at $1650.71 and here we are at >$2000 so if the shorts are not conceding it is about $24B now. The next report will be for August 31st, fortuitous from my perspective as it will put a clear number on the shorting level post-split, though we will have to wait for September 10th to find out what that is.
To put things into perspective for those who put any faith in @ihors3, he reported that the short interest had continued to decline rather than increase with an error of >13%. Last year when the stock price bottomed out he claimed it was all weak longs selling and that the shorts were covering. In reality it was shorts piling on at the bottom. Even so, his error was only 10.4% (which he made up for by then claiming that short interest was increasing when they were in fact covering, giving him a 12.8% error in the other direction).
He has been even more wrong, his winning figure was just under 19% error Feb 14, 2020 when he was still claiming short covering was substantially less than it really was. This was during the run up toward $1000. Since December 14th, 2018 (the oldest data I have) the first standard deviation for his error is 7.18%, but even more significant is how consistently wrong he is in the direction of the change in short interest.
Most likely this is simply revealing limitations in his model and data sources rather than anything nefarious on his part, but it does illustrate the peril in relying on his claims.