It enhances risk. The latest tool language and/or framework is uncharted territory
A tried-and-true solution has had a lot of bugs and issues documented and some of them solved.
It also mean the team must take a break from development to learn whatever the new thing is. Some will need more time than other, which may slow it down even further.
A lot of systems are developed in domains where "go fast and break things", A/B testing etc. would not be at all advisable.
The main selfish focus on "adopt whatever is hip this week" is to pad resumes with it, consultant can come in to help train people $$$. Consultants will fight for it to use the newest and greatest since it may increased development time $$$.
An older tech stack doesn't mean never updating the system. either. Perl still gets updates. Linux distros still get updates. In fact, often a newer piece of software will introduce bugs including security bugs that were fixed in older software long ago. All things being equal, though, using a language with better string handling and memory safety can be a boon if your old stuff is in something like C.
It's an interesting dynamic. I've been on both sides in my career. Silent is much easier but doesn't get you recognition & promotions.
...and don't forget to include the wonderboy/girl from the overpaid consulting firms that come in and tell the CEO they are doing everything wrong and should move everything to: (pick the latest technology fad)
and then 2-3 years later they come back and convince upper management to move to the next big thing...
But there's an implication of moral purity at play. Are those who just quietly "mind their own business" a good, wholesome but misrepresented group?
On the contrary, I think that in matters with far reaching consequences, which computing is, silent majorities are the problem. Unreflectively going along with the programme becomes a kind of passive negligence which should not be lionised in a folksy way.