The server will run at 100% capacity so as soon as it frees up from one task it’ll do the next.
As such, once you hit 100% capacity that means users will often have to wait longer for pages to load, data to get saved, etc. because the server isn’t immediately ready for their request but will instead put it in a queue.
Additionally, at 100% capacity, my understanding is API calls are deprioritized (and won’t run until the server is below 100% capacity again) so they often don’t run in the same order as you’re accustomed to them running relative to other workflows, which can work fine so long as your app is designed to handle this. Furthermore, I believe user’s requests can time out if it takes more than 30? seconds for your server to reply to them – but I’m not quite sure on the details here.
So, barely hitting 100% for a few seconds doesn’t really matter at all. If you have 2x more people using your app than your server can handle at 100% capacity, then it’ll be really slow for those people because tasks will queue and become really long (likely until enough people get tired of it being so slow that they drop off your site).