When a conference gathers more money (e.g., from sponsorships, registration fees, or public subsidies) than what was spent for organizing the conference, where does the conference budget surplus go?
Here are some models I am aware of; combinations of these also exist:
There is a sponsor (e.g. government, local university or some academic organization) that covers part of the expenses but their rules for sponsoring are such that you cannot make surplus. You will need to return some of the sponsor money back to them so that the bottom line is non-positive.
There is a contingency fund for the conference series. Surplus from one year is used to cover deficit in another year, or to support other conference-related activities in the future years (say, sponsoring student travel).
There is a scientific organization (e.g. ACM) that runs the conference series and takes any surplus (and promises to help if something goes wrong).
The local university that organizes the event will keep any surplus (and they will also be responsible for helping if something goes wrong).
The local organizers will figure out some way to avoid surplus. More free coffee, additional student travel grants, last-minute conference fee discounts, additional awards, free city tour, honorariums for keynote speakers.
Some part of the income can be flexible. For example, the registration fee for local participants may be fixed only after the conference, once all other income and expenses are known, so that the bottom line will be close to zero if possible.
In many (most?) fields, conferences have sponsors. In my field (CS) a typical sponsor might be the ACM (Association for Computing Machinery) or one of its parts. Any surplus would go back to the sponsor.
But the sponsor is also responsible for any shortfalls that occur.
I've never been a conference chair (only lower level positions) and never the Finance Chair, but I suspect that most conferences are expected to at least break even and maybe return a bit to the sponsoring organization. I don't have examples of "cash cow" academic conferences, but suspect that they exist. There have been some monster conferences, though they are more like expositions.
My recollection of ACM conferences is that the conference committee always has a member who represents the organization directly, looking out for lots of things and giving advice to the chair(s) if they aren't already experienced.
I'll also note that the committee is (ACM, again) composed of volunteers, though some support clerical staff might be paid. In any case, it isn't the committee that benefits financially.