seemingly uneven trade suggested by mr. stockfish

1k3r1r/1pbb4/p3pQ1p/2P3p1/5P2/8/PP4PP/R2R2K1 w - - 1 23

stockfish wants to sac the queen and a pawn for a rook and a bishop
moving the queen backwards to c3 results in a better engine evaluation as well as having more material
is that one tempo loss that white suffers from moving the queen backwards really worth 2 pawns? blacks king doesnt seem too stable after all so why get rid of the queen? is the potential pressure on g2 really that big?
so many questions

or am i missing something else?

Cliff notes version: After Qc3 Bc6 white is still winning, but black gets a lot of play, and white is nearly forced to concede material anyway. The position after the trade of R+B for Q+P is easily winning, with no real play for black.

Longer version:

Put in general terms, yes, the reason it likes trading Q+P for R+B is that what you give up there is less than what you give up by playing Qc3.

In other words, yes, black's activity, piece coordination, and kingside play after Qc3 Bc6 is worth more than the difference between Q+P and R+B.

Having said that, both are quite winning for White; I would still prefer the simpler approach with Qxh8. It's a lot harder to accidentally blunder there than after Qc3 Bc6.

Also, you'll probably want to give up the f4 pawn anyway after Qc3 Bc6.

Playing fxg5 hxg5 gives black a lot of threats on the kingside, and the win can quickly become very difficult for white with just a couple inaccuracies.

Cheers!

thx for the reassuring :)
the weird thing is that the engine does what a human would probably expect not to be done by an engine: it tries to shut down "potential" counterplay
usually thats a very human approach