[tyleo]

One of the troubles with games is that you don't know if your product will be successful up front. You can go down the route of building everything (engine, multiplayer server, game DB), but if you frontload the tech and your game isn't fun 5 years out, you're in a bad spot.

My experience has been that it's critical to frontload "fun discovery" which means taking concessions on 3rd-party technology. If you make something successful you earn the chance to replace 3rd parties with custom solutions. Often they're fine if your margins are high enough. If you fail, you won't be paying those monthly subscriptions long anyways.