It's really not "huge compensation" until they've scaled their organization to thousands of employees around the world, publishing thousands of journals & tens of thousands of books. On a per-unit basis, their profits are pretty modest.
250,000 articles, 2.7 bn in revenues, of which 1 bn is profit - that means each article generates $10,800 in revenue, which means there's a breakdown of $4000 in 'profit' from each article, and $6,800 in 'expenses,' assuming all revenues come from publication activities. It costs money to manage and publish these articles, and you don't do away with that cost by getting mad at Elsevier. If you want everything to be "free to anybody who wants a copy," you have two choices:
1) Create a federal agency that does the job Elsevier does, funded with taxpayer money, which isn't trying to earn a profit, and mandate that all taxpayer funded scientific research must be published through that federal agency;
2) Mandate that all grant money MUST publish to "some open access" platform, and make that a condition of the grant award.
If you do #1, you've created what's almost certain to be a politicized, inefficient government bureaucracy which will arguably find a way to simply cost more than the 2.7 bn in revenues Elsevier takes in, and you've also essentially "nationalized" Elsevier by legislating them out of existence, because as others have pointed out... there's a massive amount of research that's funded by taxes these days.
If you do #2, well, the situation remains the same as it is today - Elsevier will still be a for-profit agency charging an average of $10,800 per paper to publish, and researchers will just ask for a little more money to cover their anticipated publishing costs.
Really, this isn't exactly "fuck you" money that's being gouged out of every researcher. I'm not sure I think either solution is an improvement, but I'd favor #2 if it came down to it. On a per-paper basis, Elsevier isn't exactly making "fuck you" money, I'm not sure that getting mad at them (instead of the government, for not mandating Open Access publication) makes sense.