Presumably that's how many users they had 2 weeks ago, as indicated by the giant "Recent" indicator and the dates? You can always switch to "All Data".
Not zero indexing is misleading if you are comparing discrete things like GPU performance, not in the case of plotting a timeline graph. Their published stats could be seen as misleading if they only displayed a short and/or a specific timeline (excluding the latest data for example).
Hint: This has nothing to do with a zero y-axis, why would one need to have a zero value to have the y-axis start at zero?
The reason to have a non-zero y-axis for time series is to amplify changes, e.g. the changes might be to small to see with a zeroed y-axis. Or you have ups and downs and want to compare them, with a zeroed y-axis again the changes might be too small to compare.
Whenever you want to show growth, a non-zero y-axis is usually a sign that the aim is to overstate growth, because we as humans estimate growth by the steepness of the graph, not by the numbers. A non-zero y-axis creates a much steeper graph and thus growth is perceived much higher than it is.
2.) Switching to "All Data" also doesn't set the y-axis to zero, but to 6,840.
Having an y-axis set not to zero is in the majority a sign of people who want to inflate growth.