Sure. 100T/day * 1day/86400sec ~= 1B/sec. They're probably considering at least a few hundred candidates per impression, and every impression is going to go through _at least_ two models (relevance and pCTR/revenue), so you could get there just with online serving at 5Mqps, which is plausible. But they're also going to be doing a lot of stuff in batch - spam predictions, ad budget forecasts, etc - so that every candidate actually runs through four or five different models, and every actual impression could do more than that.
How many ads does Meta serve a day, and how many AI model executions are done for each one? Repeat the same for stories, post and comment recommendations on Facebook and Instagram, and you have very big numbers. To that, Add VR, internal modeling and other backoffice/ offline analyses over billions of users and you'll easily get into the trillions.
Perhaps there's some combinatorics where every time an ad or post is displayed to the user, it runs through some hundreds/thousands of candidates and computes their relevance.
Such a large number, makes sense?