# The 'Never Write Your Own Engine' Rule Is Quietly Dying > Source: > Published: 2026-08-14 23:08:39+00:00 [Dev Tools](https://sourcefeed.dev/c/dev-tools)Article # The 'Never Write Your Own Engine' Rule Is Quietly Dying Unity's trust collapse, mature low-level libraries, and LLM scaffolding have rewritten gamedev's oldest advice for a narrow class of studios. [Lenn Voss](https://sourcefeed.dev/u/lennart_voss) Custom game engines are supposed to be dead. According to Video Game Insights' engine report, only 13% of games released on Steam in 2024 ran on one, down from 71% in 2012. [Unity](https://unity.com) and [Unreal](https://www.unrealengine.com) won; the "make a game, not an engine" lecture became gamedev's most repeated advice. And yet those 13% of releases still pulled in roughly 40% of Steam's revenue — and a steady trickle of small studios keeps opting out of commercial engines entirely. The latest is [Team KwaKwa](https://team-kwakwa.com/soup-raiders), the small Lausanne-based studio behind Soup Raiders, a tactical RPG about pirate animals fighting inside a bowl of soup. Co-founder Elias Farhan — who also teaches games programming at SAE Institute Geneva — is porting the game off Unity onto a hand-built C++ engine, targeting 60 Hz on Nintendo Switch with sub-three-second load times, and documenting the work publicly. The interesting question isn't whether one small studio can pull this off. It's why the standard advice against trying has quietly stopped being true for a specific class of developer. ## The risk calculus flipped The old argument against custom engines was about opportunity cost: every month on a renderer is a month not spent on your game. That's still true. What changed is the other side of the ledger — the risk of *depending* on someone else's engine. Unity's September 2023 runtime fee, a per-install charge applied retroactively to shipped games, was cancelled a year later after new CEO Matt Bromberg took over, but the damage was structural: developers learned their business model could be rewritten under them mid-project. Unity has also revoked individual licenses with little recourse — locking developers out of the editor and, in some documented forum cases, their own work. And engine vendors can simply vanish. When Our Machinery shut down The Machinery engine in August 2022, licensees weren't just abandoned; they were instructed to delete all copies of the source and binaries. Any game built on it lost most of its accumulated value overnight. Veterans remember RenderWare — the "Unreal of the PS2 era," powering GTA III through San Andreas — evaporating as middleware after EA bought Criterion. The revealed preference of indie developers says they noticed. At this year's GMTK Game Jam, [Godot](https://godotengine.org) ended Unity's nine-year streak as the most-used engine — 47% of submissions to Unity's 34%. Most of those developers fled to another engine, not away from engines. But the same trust collapse that drove the Godot migration is what puts building your own back on the table: both are bets on code nobody can take away from you. ## "Custom engine" no longer means "from scratch" The phrase conjures a decade of yak-shaving before you render a sprite. That's not what shipping-quality custom engines look like in 2026. [SDL3](https://libsdl.org) handles windowing, input, audio, and its new GPU API abstracts Vulkan, D3D12, and Metal. Dear ImGui gives you tooling UI in an afternoon. FMOD or miniaudio covers sound. A "custom engine" today is a curated stack of battle-tested C and C++ libraries plus the one layer that actually matters: a game loop and data model shaped exactly like your game. That specialization is where the wins live, and they're bigger than they look from inside Unity. Factorio's developers have said flatly that a standard engine would leave too much performance on the table for their simulation. Animal Well shipped as a ~35 MB executable with a full-screen fluid simulation because Billy Basso's engine does only what Animal Well needs. Dan Baker's Nitrous engine does a clean build in under two minutes and runs debug builds at 30 fps — compare that with Unity IL2CPP builds that developers routinely clock at over an hour. Iteration speed is a compounding asset, and it's the one thing you can't buy from an asset store. There's a rendering argument too. Unity and Unreal's hardware interfaces were designed in the DX11 era and had to emulate that model on top of Vulkan and DX12, eating CPU time in translation layers. An engine designed around modern explicit APIs from day one skips that tax entirely — ex-Unity graphics veteran Sebastian Aaltonen has argued this for years, and now adds that LLMs writing the generic scaffolding remove the last excuse: you focus on the code that differentiates your game. ## Who should actually do this Be honest about the 13%, though: the default advice survives because it's usually right. A custom engine still makes no sense for most teams, and the failure mode — years of engine work, no game, burnout — is well documented. The console problem alone is disqualifying for many: Unity's porting story genuinely is magic, and doing a Switch or PS5 port by hand means NDAs, devkits, and months of platform plumbing that Farhan's Switch targets will have to prove out. The profile where it works is narrow but real: a 2D or stylized-3D game with unusual mechanics or performance needs, a programmer-led team that has shipped before, one or two target platforms, and a genuine allergy to per-seat fees ($2,310/year for Unity Pro) or royalties (Unreal's 5% after $1M). If that's not you, Godot — MIT-licensed, no vendor to revoke anything — captures most of the independence upside at a fraction of the cost. And if it is you, one rule from the people who've done it keeps recurring, in Tyler Glaiel's formulation: make the game *at the same time* as the engine. An engine validated against a real shipping game stays a means; one built in a vacuum becomes a very elaborate way to never ship. Soup Raiders already exists as a game — which is exactly why its native port is worth watching rather than dismissing as another engine-programmer's dream. ## Sources & further reading - [Soup Raiders Goes Native: What You Gain by Building Your Own Game Engine](https://eliasfarhan.ch/gamedev/cpp/2026/08/14/srnative-01-why-a-custom-engine.html)— eliasfarhan.ch - [The Big Game Engines Report of 2025](https://vginsights.com/assets/reports/The_Big_Game_Engines_Report_of_2025.pdf)— vginsights.com - [Godot has become the most popular engine at GMTK Game Jam 2026](https://gameworldobserver.com/2026/07/29/unity-is-no-longer-the-leader-godot-has-become-the-most-popular-engine-at-gmtk-game-jam-2026)— gameworldobserver.com - [Unity is Canceling the Runtime Fee](https://unity.com/blog/unity-is-canceling-the-runtime-fee)— unity.com - [The Machinery game engine cancelled as developers told to delete all source code](https://www.nme.com/news/gaming-news/the-machinery-game-engine-cancelled-as-developers-told-to-delete-all-source-code-3283633)— nme.com - [Soup Raiders](https://team-kwakwa.com/soup-raiders)— team-kwakwa.com [Lenn Voss](https://sourcefeed.dev/u/lennart_voss)· Cloud & Infrastructure Writer Lenn writes about cloud platforms, Kubernetes internals, and the infrastructure decisions that quietly make or break engineering organizations. Based in Berlin's vibrant tech scene, they have a talent for turning dense platform-engineering topics into prose that people actually finish reading. ## Discussion 0 No comments yet Be the first to weigh in.