Money not being an issue can still happen: See Valve, which have a money printing machine in Steam, along with a few long living, lucrative games. Work on Half Life 3 never really stopped , and we had a 13 year wait between Half Life Alyx and Half Life 2, episode 2. Sources say that the internal standards for a Half Life 3 are so high, it led to many restarts and a lot of work being thrown away.
But the real reason we don't have games that run as long as Deus Ex is that the cost of reworking what was already there have exploded since those days. If you are predicting a 3 year release cycle(not unheard of to this day at large publishers), and you are delayed by 2 extra years, the hardware you are targeting is different, and the amount of art you have to upgrade or completely remake is crazy compared to Deus Ex. With teams being far larger than they were to meet a 3 year plan, marshaling a modern AAA game team to rework large parts of a game has little to do with managing to do the same with a team from the late 20th century. A linear increase in team size gets you a geometric increase in coordination costs.
That said, we still have small teams that might work on a game for a very long time, but they have to aim far lower technically, or in scope. You can make a beautiful, shorter game with a small team, like Outer Wilds. You can also make a 2d game, and tweak your game to perfection, like Stardew Valley. At those sizes, saying "when it's done", and letting yourself change direction and scope is fine. But try that with the Red Dead Redemption 2 team, trying to be at the forefront of everything. Without a far more rigid plan, you can go into a Duke Nukem Forever spiral (and DNF was a far smaller team!)
An important note with specifically Stardew Valley and a lot of the indie games coming out nowadays is that they release early; they work for up to a year, then release the game in Early Access to start the flow of money coming in so that they can both keep working on the game and get a lot of instant feedback.
I don't think that approach would work well with story-driven games, but for e.g. Stardew Valley and a lot of the infinite gameplay indie games being made nowadays it's ideal. Also, they will add features all the time, so people will come back to them on a regular basis (e.g. Factorio, KSP).
Game graphics standard were changing much more rapidly back then. Games released in 1995 and 2000 look much more different than games from 2015 and 2020.
Couldn't a few artists and level designers in 1995-2000 crank out a whole lot more finished products than ones in 2015-2020 though? Seems like you need entire teams to make one level in modern games.
In 1995 individuals would crank out doom WADs and marathon maps like it was nothing. 2.5D was a sweet spot for ease of making something. As a tween with no artistic skill and the family computer I could make maps and mods and texture stuff.
Things got more difficult when everything became legitimate 3D because of unreal and quake. It still happened but IMO was a much steeper curve.
That is my point, coordination costs are commonly modeled quadratically - as the number of edges in a fully-connected graph being a quadratic function of the number of nodes.
But the real reason we don't have games that run as long as Deus Ex is that the cost of reworking what was already there have exploded since those days. If you are predicting a 3 year release cycle(not unheard of to this day at large publishers), and you are delayed by 2 extra years, the hardware you are targeting is different, and the amount of art you have to upgrade or completely remake is crazy compared to Deus Ex. With teams being far larger than they were to meet a 3 year plan, marshaling a modern AAA game team to rework large parts of a game has little to do with managing to do the same with a team from the late 20th century. A linear increase in team size gets you a geometric increase in coordination costs.
That said, we still have small teams that might work on a game for a very long time, but they have to aim far lower technically, or in scope. You can make a beautiful, shorter game with a small team, like Outer Wilds. You can also make a 2d game, and tweak your game to perfection, like Stardew Valley. At those sizes, saying "when it's done", and letting yourself change direction and scope is fine. But try that with the Red Dead Redemption 2 team, trying to be at the forefront of everything. Without a far more rigid plan, you can go into a Duke Nukem Forever spiral (and DNF was a far smaller team!)