Just a very rough calculation: If Nature has a staff of around 150 [1], each paid $100,000 year, 50 issues per year, 20 research articles per issue. This gives
150 times 100,000/(50 times 20) = $15,000 per article.
My numbers are probably off in more than one way.
https://news.ycombinator.com/item?id=16339193
Nature drops about 95% of submitted papers somewhere in the pipeline, so part of the costs applies to 20x more papers than you see in the end.
If an editor can survive with these fees even when the paper will be publicly available, how can Nature need $30,000 for a paper people will pay to read?
Ordinary scientific journals don't have reviewers in paid staff. An editor sends an article (without reading it) to 2-4 non-paid reviewers, and makes the publish / don't publish decision based on the reviews. Also the editor is not paid, either.
Also, vetting submissions should be a relatively small part of your time. When you're reading a lot of submissions, you learn how to weed out the weak fairly quickly.