Most games are built on engines their developers did not write. The decision is straightforward: an engine is a large software project whose completion produces no game at all.
An engine is infrastructure, not content
Rendering, physics, audio, animation systems, networking and asset pipelines have to exist before anyone can build a level. None of that is visible to a player as such.
Writing them takes specialist engineers years, and the work must be maintained continuously as hardware and platform requirements change.
A studio that builds its own engine is funding a software company alongside a game studio, which only makes sense at considerable scale. The cost is justified across many projects, not one.
Tools matter more than raw capability
What a licensed engine actually supplies is a working toolchain: editors, importers, profilers and build systems that let designers and artists iterate without an engineer.
That iteration speed determines how many versions of a level can be tried, and trying more versions is most of how games become good.
An in-house engine with weak tools slows every department, which is a common and expensive failure that is invisible from outside the studio. Players see the result as a game that feels underdeveloped.
Hiring is easier against a known engine
Developers experienced with a widely used engine can be productive quickly, while a proprietary engine requires months of internal training for each hire.
That matters in an industry with project-based staffing, where teams expand for production and contract afterward.
It also reduces risk when key engineers leave, since the knowledge they hold is not unique to one company's undocumented systems.
The costs are licensing terms and constraints
Commercial engines are paid for through fees or a share of revenue, which is a permanent claim on the game's income rather than a fixed development cost.
An engine also embeds assumptions, and a project working against those assumptions spends its time fighting the tool instead of using it.
Studios with unusual technical requirements, such as very large simulated worlds, are the ones that most often still build their own.
Custom engines remain where a game is the engine
Some studios treat their technology as the distinguishing feature, where destructible environments or particular simulation behavior is the reason the game exists.
In those cases the engine is not infrastructure but the product, and licensing would mean giving up the thing that differentiates the work.
Everyone else spends the saved years on design, art and content, which is what players are ultimately buying. The engine question is settled by what a studio wants to be distinctive at.