Because large urban areas take a long time to count ALL of the votes. Yea, where a lot of black people happen to reside in. Doesn't make it the bit suspicious.
Funny, in that it works the opposite way here. Every single election I can remember staying up to watch the count for (going right back to childhood) one would see Labour appear to take an 'early lead', only to go to bed and wake up to find the Tories ultimately won, or at least pulled things back a long way, overnight.
Because, kind of obviously, the densely populated urban areas that tend to vote Labour have shorter distances so get the ballot boxes to the counting stations much more quickly, while Tory-voting areas tend to be rural, thinly-populated, and spread-out, so it takes much longer to gather up the votes. So the early results are all Labour seats, while the Tory seats tend to come in later in the night.
(I'm also tempted to suggest rural people just do _everything_ more slowly)
In the US 2020 vote I would assume things played-out quite differently to whatever the normal pattern is in the US because of the unusually heavy use of postal-voting. But there's absolutely no reason to expect the first votes counted to be predictive of the final result. It's not going to be a random-sampling of the total vote.
If the pattern seeming to change as more votes are counted is presumed to be 'suspicious' then I could claim every single UK election ever was 'stolen' by the Tories.