I am going to look at the negatives for Foundry and FGU. I have no significant time with roll20 as a GM so I can't speak about it.
I think that the core problem with Foundry is that it is still in beta. Changes will happen because more features are implemented. The steady stream of upgrades has created burn-out for extension developers and extensions can get wildly out-dated as a result. For a 5e player, a further complication is that the dev for the 5e ruleset is also the Foundry developer and as a consequence, the 5e ruleset is well behind several others (WFRP in particular, is excellent).
The negatives for FGU have largely been echoed previously in this thread. They should have improved the UI - it's a no-frills trip back to 2003 that makes GMs spend too much time telling players how to accomplish simple tasks when we should be gaming. It's also a bit behind on visual gloss (animation) and not even in the game for sound effects. Also, a lot of the older modules are very low quality in terms of graphics. I could really rant here, but I will simplify it by saying that Smite should have minimal standards for what is acceptable for a module and should sweep out some of the aging products that offer very little of value to a GM.
POSITIVES... let's end on a good note.
Foundry has the ability to be hosted on a server and hence even GMs with poor connections can host visually rich campaigns. There are a good number of free extensions / rulesets. The multimedia support is outstanding. Development speed is noticeable and it is continually improving.
FGU has the best automation of well supported rulesets. (That's mostly 5e, 2e, and Savage Worlds - hopefully I've not missed one.) It has superb support for effects. The community here is amazing and you have to credit Smite with having made mostly good choices in the management of the community.
MOVING FORWARD...
I think that Foundry is ahead in the areas that they've worked upon. They will get effects in soon and once all the individual elements are built, they'll dig in on automation. They will close the gap as they reach 1.0.
FGU has a great community and that might be really beneficial to them as competition heats up. Smite needs to get sound added, they need to get animation going, and they need to fix-up their difficult interface. Since they have an actual software client, in my opinion they should try to link in a PDF viewer. If they added PDF support and used PDF notes for hiding data, they could let all of us spend NO TIME entering text while having it right there as the authors intended. (PDF support is purchasable as a support library specifically for Unity, so this is far from an impossible dream.) I would also love it if Smite tried to work a deal with Syrinscape for a sound engine - let it play / work without a subscription to Syrinscape, but make it super-easy and even more feature rich with a sub. Smite should take advantage of being a software solution and do the things that a web solution cannot - that is how they find and maintain a competitive edge.
Top

