It's not. The (in)famous epi.org is flawed for all sorts of reasons, from excluding noproduction/supervisory workers (the highest compensated ones!), to excluding non-wage compensation (eg. benefits), to different deflators for compensation vs productivity. If you adjust for all of that, the chart is unremarkable.

https://www.piie.com/blogs/realtime-economic-issues-watch/gr...

That incidentally, is the exact problem with the site. It presents a barrage of charts, without regard to relevance or rigor, and tries to persuade through sheer volume alone. Yet, if you scrutinize any of them, it quickly falls apart. That's probably why the site doesn't even bother justifying the charts, or even state the thesis, for that matter.