> Excel is one of the most difficult software to replace in a corporation. It just get the work done, without any third party to block the process.
I think this is a common impression which is both a result of poor corporate IT governance/process (and, largely, the fact that "IT" is usually a giant silo that is separated from the business as if it were an outside firm, with rules that only those in that silo can touch anything that might be used to create functionality -- except Excel) and, even so, largely illusory. (There's no less need for systematic testing with Excel, but the people using it often don't understand that, and build complex, poorly-tested, bug-ridden systems that look authoritative.)
> Ok, you can do it in a better way (faster, online, distributed, ...), but it creates a lot of costs too.
More important, it creates costs that are part of the business units expenses, but where the work (and headcount) is elsewhere, meaning that the costs of Excel -- even if they are equal costs -- are often desirable. While managers usually have incentives to cut costs, empire-builders were prefer that, if there are costs that are unavoidable, the associated staff are theirs, not someone else's.
> In a large corporation, it is not just about the money, they'll waste a lot of time just to define what they want and who they want to make it happen.
Yes, Excel is often a loophole in horribly broken enterprise IT practices that function more to create bureaucratic overhead and prevent IT work from getting done than to serve the needs of the organization, and rapidly accumulating technical debt by having business staff build complex tools in Excel without consideration of whether its an appropriate tool for the business need outside of the role of Excel as an organizational IT policy hack, without structured testing, etc.
OTOH, fixing the broken practices is a better solution.