The sole reason while your research generates millions of dollars of year of saving is because it would be applied at Google scale. So your work times google scale = $10M/year.
As stated above, you deserve to be compensated for a part of your work. But pretending that your compensation should also include the "Google scale" factor - which you did not contribute to in any meaningful way - is not right.
If you want your compensation to be based on the outcome of your work on the company as a whole, you need to work for equity and not for salary, and accept the associated risks - which might include getting nothing for a lot of efforts. Of course, your compensation will then also include the work made by others, which you might like or dislike.
If you want to be able to precisely measure your own contribution, exclude any other factors, and capture almost all of it, you probably are better of creating your own solo-founder type of company and stay solo. When 1+1=3, it's hard to precisely pinpoint which "1" generated the extra value.
Google probably runs at 25% profit margin, so $10M/y savings equate to $40M/y additional revenue. A Google cloud sales rep probably makes at least 5% sales commission, so they'd net at least $2M for equivalent sales (which they would only able to achieve because of Google scale, but if you make that argument with them, good luck keeping them employed). Conversely, a SWE delivering $10M/y savings would probably only see a small bump on their refreshers and bonus.
"Google scale" wouldn't exist without the workers that make it possible. Seems to me like they can get more of that value they enable than they currently do.
If I can make $1 on my own, and you can make $1 on your own, but together we can make $4, we each deserve $2. I don't deserve $3, even though I turned your lone $1 into $4. But I also don't deserve just $1, because I did contribute to the scale factor. In todays world, it seems like we'd each get $1.
How do we divide the extra $100k. Are you pretending that you should get $50k (50 thousands times what you were doing alone) and we should get $50k (1.005 times what we were doing without you) as well ? That does not seem very fair to me at all.
This isn't a good model. Boosting your company's profit by 100k doesn't mean they can boost any company's profit by 100k. If they could, they'd capture a lot more of the 100k.
In theory, if you had a competitive and fungible marketplace where a specific employee would boost profits by $100k, and you offered her a salary of $80k, someone else would outbid you with $90k, and someone else would outbid them at $95k etc. Even if the "margins" are low, no one is going to pass up on free money.
In practice, the labor marketplace isn't nearly competitive or fungible enough to expect the above. However, there is still a very different problem. Even if the average salary is "fair", people get rewarded a tiny fraction of the value they generate. Imagine if I'm working at Google and I find a way to save Google $10M/year via improved hardware efficiency. Despite the fact that this isn't part of my job description. How much of that $10M would I get in the form of a salary increase or bonus? A pretty darn tiny amount. 1% if I'm lucky. Is it worth it? Probably not. I might as well just do what my manager asks me to do, and spend my extra time learning Haskell and going to happy hours. This is pretty much what GP was alluding to. If companies started giving people a significant fraction of the value they generate, corporate life would change dramatically.