It's mostly restrictive on the side of game design. Once your goal is to support a large quantity of software features you automatically fall into the mold of a style of gameplay that will minimize impact on feature development.
In the classical model of ship-once gamedev, certain big-ticket software features tend to be thrown out when the designers are in control, because they're so costly to get right and don't surface as much immediate impact:
* Large-scale persistence(just store a minimal representation and instance that)
* Online features(too much synchronization logic - it hinders almost every other feature!)
* Complex AI and procedural simulation(too much chaotic behavior, limited ability to control outcomes)
The transition we've made has, in essence, been to construct microtransaction service products that do support these more complex features: the game is a platform for the sale of assets, and it is massively online with total persistence and sometimes even robust AI. And when you start doing that, your ability to revise the design diminishes substantially. For the most part, you can't take away assets and features that are already there without a substantial player backlash, which means that games of today tend towards accumulating their way into an incoherent maximalist experience with no particular focus or intended playstyle, all of it following templates intended to encourage monetization:
Make sure the players can customize their character cosmetically, and give them a huge array of skills and powers, but nerf all the unique aspects of these powers so that the game will play roughly the same regardless of what you choose. Present them with opportunities to draw in their friends, and add hooks and incentives to come back repeatedly and to check the in-game store for deals. Provide play opportunities for all segments and demographics of players(following the lines of the Bartle player types).
When this model works, it does extremely well in the mainstream(Fortnite basically ticks all the boxes and is outstanding in its ability to progress as a live service) - but it has the downside of being a compromise for everyone. It makes the game too Byzantine in its complexity, too restrictive in its customization, too carefree in its changes to competitive dynamics, too willing to let annoying minor bugs persist - to say nothing of the cost and complexity needed to pull it off.
So I think there's definitely room for more of a spectrum of products, from the mega-blockbuster down to tiny boutique indie games. But the marketplace is still catching up with the idea of providing full service, and a lot of games are still released with the focus being on developing a franchise or on addressing a specific genre. In many cases the capital requirements needed to do services crush that idea out of the gate.
(I'm guessing the biggest factor is the massive investment required to make a game nowadays.)