Some of the most important parts of a host are not even hardware related. It would be difficult to find a datacenter that offers machines incapable of running an OT.
Things you should consider:
RAM is usually expensive, shop around for the amount of ram you think your server will use. I would overcompensate here but not too much due to price. Keep an eye on how much memory your server is using as you gain players and upgrade as you need it. For reference, my server (Deathzot) has 32gb of ram (because it was cheap as hell at OVH) and I cannot recall a single time it has used anywhere near that, 16 probably would have been more than enough. I will say, I've never ran a real map so I'm not sure how much memory it uses but I assume it isn't that big of a difference.
Not sure if this will help at all but currently Deathzot (based on TFS 0.4) is using around 1.5gb with ~10 players online, for comparison my TFS 1.x test server has around 5 players online testing and uses <100mb so maybe those numbers will help you decide what you need.
Location is extremely important, even on the best datacenter in the world, if your playerbase is in the US and the server is in China, its going to feel like shit.
Service... This is so hard to judge before buying and it is so important. How often are outages, how quick are they to respond to issues, how strong is their network, what is their policy for DDOS attacks. The last thing you want is for your server to be down for hours every week because they are constantly having issues, or them shutting your server down because its being DDOSed and they don't want to deal with it.
Something to keep in mind... Your server (if long term) will have hardware failures, it will be DDOSed, and it will experience outages due to your host having issues. The question is how big of an impact will these things have overall. A good company will be able to get your server back up and running extremely fast even under the worst conditions.
A good example, OVH Montreal, had a person run into a power pole and it punctured their trunk line between Montreal and USA, they had a backup line functional within hours (tho poor performance) and the main line was mostly fixed a day later, I think the complete repair took a couple days total but most of their service was back online within 24 hours.
A bad example, we actually had a company email us and tell us that we needed to go and purchase 3rd party DDOS solutions or they would turn our server off. In addition, every time the server was hit with an attack it was turned off for around 24 hours, even if the attack only lasted a few minutes.