Instead, you now put money and time in building a wheel for very little benefit (if you can feel the performance of jQuery, you're doing to much DOM operations and you probably wouldn't benefit much from stripping it out).
Some projects are not meant to last long-term, and jQuery would work just fine. Even if it is old fashioned.
Can you elaborate on the runtime overhead issue? If you write your own JS functions to replace jQuery, you’re still adding runtime overhead... and jQuery likely wrote a more efficient implementation to begin with; jQuery is battle tested and one of the most widely used pieces of software in history, what makes you think your code will have less runtime overhead?
Have you ever measured this on a large scale? I’ve never found it to be true because there are many CDNs, versions, and browser caches are not infinite. If performance is a priority it’s almost always faster to serve it directly to avoid things like the DNS and TLS latency from those CDN connections.
All this talk of trying to skim off a few kb of load time only applies to extreme edge cases of landing pages with flighty potential customers.
And your suggestion is write your own custom wrappers to avoid overhead? So you or someone else can learn all your custom things.
No to all of that. Get off the "too cool for Dad" train and just use jQuery if you already know it and it fits for what you are doing. If something else is better, (including native), use that.