Very few new games use their own engine. Excluding legacy titles and AAA publishers who have their own tech developed across decades, I believe proprietary engines are less than 10% of the market today. The only exception is the Japanese market but even that has changed in recent years with Unity and Epic making inroads.
It’s very difficult to justify the expense of developing your own engine, especially if you intend to release for multiple platforms.
It’s very difficult to justify the expense of developing your own engine now. Those numbers are distorted by Unity having been on the market for seventeen years. In 2005 (the year Unity hit the market), there were thirty-nine new games on Steam. In 2021, there were over 10,000.
Now you have to ask how many of those games would have happened absent a Unity, and the answer is "Most would not." But it's a chicken-egg problem: without Unity, the games wouldn't have existed, but if Unity asks for too much money to publish the game with their engine, the game also doesn't exist.
It's a conundrum that caused Unity's predecessors to mostly try and fail (with a few exceptions that are still around), and I've been impressed by their ability to thread the needle on this market. But it's a delicate market... lacking a Unity out there, I doubt OnlyCans Team would have rolled their own engine, but I also doubt they'd have paid money to Unreal to execute on their novelty idea. They just... Wouldn't exist. Unity has to pull far, far less revenue from developers of a game project than Unreal does to maintain the existence of the ecosystem they grew around themselves.
> In 2005 (the year Unity hit the market), there were thirty-nine new games on Steam. In 2021, there were over 10,000.
While I think your point still holds, I think it should be noted that 2005 was 2 years after steam released. And still primarily a distribution platform for valve’s first party games. There were many more games available via other distribution channels.
It's also worth noting that Steam has partially opened up for independent developers in 2012 with Steam Greenlight (via manual curation), and fully opened up only in 2017 with Steam Direct. Before that, you had to be an approved partner to be able to publish on Steam.
It's unlikely for a small indie to roll a mobile game engine But if you are talking about companies with millions of dollars of income from gacha games like Hearthstone, most of them have already attempted to roll their own or even used in production already. (NeoX from NetEase, Cyllista from Cygames and Supercell’s unnamed engine…) For mobile game engines, they are not striving for crazy visual fidelity since mobile hardware is really limiting. Those who stick with Unity now because it’s cheaper, but if Unity tries to charge way more, they will definitely spend that markup to push their own tech onto production instead of surrendering money to Unity.
Unreal has technology moats like Lumen and Nanite to justify the royalty while Unity doesn't.
There's also a difference between rolling your own in-house engine like Frostbite or REDengine, and making a game on your own with no general-purpose engine using something like SDL. There's little point in doing the former nowadays, but there's plenty of titles that still successfully take the latter approach. General-purpose engines make this whole field obviously more approachable and for some kinds of projects are the only viable option, so the percentage share of no-engine games will likely continue going down, but I don't expect absolute numbers to drop significantly.
It’s very difficult to justify the expense of developing your own engine, especially if you intend to release for multiple platforms.