Definitely not. The engine used for BG3 is proprietory to Larian and there's no-one trained and experienced with it outside of Larian. It would be truly insane to do that when engines like Unreal 4 and 5 exist, and have huge numbers of people experienced with them. Plus it would cost a huge amount to licence the engine and pay Larian to train your staff, and WotC are penny-pinchers.
Larian have made nothing but RPGs for the entire time they've existed, apart from one RTS, which was a huge flop. And they've already hinted that both the upcoming projects are RPGs (and outright stated that neither is DOS3).
Whether it's a CRPG is harder to say - I'd take the bet and say it was except the definition can be somewhat nebulous (like, DAO is a CRPG, but what about DA2, DAI? DAV is definitely not a CRPG). They've also had insane serial success with CRPGs and seem to like them so I think it's very safe to say the next big game they make (they say the scope is less than BG3, but still large) will be either a CRPG or story RPG (i.e. BioWare-style). I think they may also be developing a tactical RPG, and that would be a lot quicker so might beat this to market, but my suspicion is that will get cancelled or already has been.
I don't think Larian have ever expressed dissatisfaction with their tools - have you seen them do so? So I am not seeing the reasoning behind assuming they are.
I'm not sure if you're referring to WotC here, but just to clarify, Sigil is an entirely separate tool (not engine) unrelated to Larian's BG3 engine in any way, shape, or form. The engine that runs BG3 is Divinity Engine 4.0, which is the latest development of a fully in-house developed original engine Larian created for Divinity: Original Sin. Sigil is a tool (again, not engine) made in Unreal Engine 5, by WotC. They might look a little similar in the scenes they create due to the subject matter and perspective, but there's zero software or asset relationship. Apologies if this is totally unnecessary and I'm just failing to follow the convo properly!
I would personally be pretty surprised if Larian switched to Epic's Unreal engine for a number of reasons:
1) Whilst the initial cost of UE5 is low, and you have to pay Epic 5% of your gross revenue for any sales over $1m gross. Given BG3 has sold like, minimum 15m copies at this point (likely far more), at $60 each, that'd mean Larian would have to pay Epic $42m. Is it likely to cost them $42m to stick with their own engine/could they save $42m by switching? That seems very unlikely (I can expand on this if people really care but it's very boring!).
2) UE developers are plentiful, but Larian doesn't like to unnecessarily fire people, nor does it operate in the ultra-capitalist "please the stockholders!" way most companies do. So hiring a bunch of experienced UE guys whilst firing their own guys would likely not work for them. And sure they could retrain all their current guys to use UE5, but why? That just adds to the tens of millions it's already going to likely cost them from the sales of a future game if they used it.
3) Epic are the sort of company that Swen doesn't seem very keen on. So I'm not sure he'd want to go out of his way to give them tens of millions. Note that Baldur's Gate 3 is not available on the Epic store.
There is one potential countervailing factor however. Both Epic and Larian are part-owned by Tencent (40% and 30% respectively), so maybe Tencent could nudge Epic on this? Or nudge both? But I'd be surprised personally. It's not impossibly but I think the odds are on them sticking with upgrading Divinity Engine - it's clearly possible to upgrade pretty impressively, given the huge leap from DOS2 to BG3 (including bringing in tons of mocap'd cutscenes!).
Ahhh yes FORTRAN used to hang around a lot in the '90s (still does a bit, mostly at older banks) - my wife's first real programming job, when she was still in uni, was to single-handedly move a hugely elaborate program written in FORTRAN 66 (not even 77 - which would still be before she was born!) to a new, modern, codebase. The program had been updated by its programmer from the 1960s to the early 1990s, but he'd retired a few years earlier, and the whole thing used variable names which were just six digits, no letters, no words. As a bonus there was no documentation of the code whatsoever, only of the user-side stuff. It had to work absolutely perfectly because it was for calculating the physical stresses on a certain kind of structure they produced, and could potentially cost the company millions if it failed. But she did it - some early code archaeology! They sensibly ran a huge number of tests too, but it worked and gave the right results! And once they had it in modern code (I forget what), she was able to add new functions, new structures, and so on, which was what they wanted, and made them god knows how much money because other similar companies didn't have flexible tools like that. Of course being a brilliant-but-naive teenager she was wildly underpaid for doing this, but she had no idea until later. Around the same time my really cool driving instructor (who taught me to do burn-outs etc.!) quit driving instructing because he knew FORTRAN (he must have been in his late 30s) and was offered an £80k job (not bad for late-90s London) updating stuff in the run-up to Y2K.