The Great Unbundling of Game Engines

For the past two decades, much of game development has been defined by a single architectural choice: pick Unity or Unreal, and build everything inside it. These two engines account for roughly 80% of all games released on Steam today. Their value proposition was integration: a single platform that handled rendering, physics, scripting, asset pipelines, deployment, and even monetization in the case of Unity. For most studios, there was no practical alternative.

That is starting to change. Jim Barksdale, the former CEO of Netscape, once said “there are only two ways to make money in business: bundling and unbundling.” The game engine market has been a story of bundling for two decades. Now, two forces are converging to unbundle it. On one end, a new generation of AI-native tools is abstracting away content creation, animation, and scripting. On the other, open-source runtimes are commoditizing the infrastructure layer that engines provide as a platform.

The Business Behind Engines

Unity and Unreal dominate game development, but neither company is primarily in the engine business. Unity’s Grow Solutions segment (advertising and monetization) generated $1.22bn in revenue in 2025, nearly double its Create Solutions segment ($621m for engine subscriptions and services). At its core, Unity is an advertising company with a game engine as a distribution channel for its ad network. When developers leave, it is not just engine subscription revenue at risk, it is the ad network and monetization flywheel downstream.

Epic’s economics tell a different story. Unreal Engine royalties are estimated at a few hundred million annually, a fraction of what Fortnite generates. Unlike Unity, Epic does not depend on Unreal adoption to drive its core business. Fortnite succeeds regardless of how many studios use Unreal. Epic uses Unreal primarily as a funnel into the Epic Games Store (EGS), where reduced royalties for on platform sales incentivize developers to distribute through Epic rather than Steam. That said, the strategy has not gained much ground yet. EGS hit $1.09bn in total player spend in 2024 (with most of this coming through its first-party IP like Fortnite, Rocket League, and Fall Guys), compared to an estimated $16bn+ annually for Steam. EGS has been unprofitable since its 2018 launch, compared to Steam, which reportedly has one of the highest profits-per head in all of tech (~$3.5m). Both Epic’s engine and store operate at a loss, subsidized by a single game. Still, the distinction matters: as a business, Unity is structurally exposed to unbundling in a way that Epic is not.

Unity is already getting squeezed from both directions. Godot is eroding its indie and lower-budget market share from below, while Unreal is capturing more AA and mid-market studios.

AI Unbundles the Top

The most commercially significant AI tools in game development are not built into Unity or Unreal. They are standalone, engine-agnostic products.

As an example, Meshy is an AI platform that converts text prompts and 2D images into production-ready 3D assets. By March 2026, it had surpassed 10m users and $30m ARR. Its latest version produces mesh generation that rivals manual sculpting, and critically, Meshy exports to any engine. Asset creation has been decoupled from the engine entirely.

This pattern repeats across the production pipeline: Inworld builds dynamic NPC behavior and voice; Scenario trains custom AI models for style-consistent concept art; Kinetix converts standard video into 3D character animations, replacing traditional motion capture workflows. None of these tools require a particular engine to function. The production layer (which can represent 60-80% of a game’s total budget) is being pulled out of the engine and into standalone tools.

Analysts at Benchmark Company (a StoneX Company) describe what is happening as a direct parallel to the microservices revolution in enterprise software: “developers can now snap together best-of-breed AI tools, such as isolated asset generators and standalone NPC behavioral models, like Lego pieces. Just as Kubernetes decoupled the application from the underlying infrastructure, AI tools are decoupling the creative process from the traditional game engine.”

A Google Cloud survey of game developers found that AI tools have compressed team requirements so dramatically that three-person teams are now producing visual output that required 30 people just a few years ago. Historically, small teams relied on the engine’s built-in tools because they could not afford to hire specialists. Now, small teams can assemble a modular production stack that fits their specific needs rather than conforming to the full weight of a monolithic platform.

Both Unity and Epic are embedding AI into their engines in response. Unity unveiled a beta at GDC that lets developers prompt casual games into existence with natural language. But developer reception has been mixed. Many developers and studio heads report that their real productivity gains come from third-party tools, not from engines’ native AI features. The AI value is accruing outside the engine, not inside it.
‍

Open-source Commoditizes the Bottom

If AI is pulling value out of the top of the engine stack, open-source is commoditizing the bottom. Gaming has been one of the last major software categories to make this transition.

There are real structural reasons game development stayed proprietary for decades. Console manufacturers (Nintendo, Sony, Microsoft) gate access to their proprietary SDKs behind NDAs, making it difficult for open-source engines to offer native console support. And the GPL license model, which requires derivative works to remain open, conflicts with how most studios protect proprietary game code and assets. Historically, proprietary middleware (such as Wwise for audio, Havok for physics, or SpeedTree for vegetation) integrates deeply with Unity and Unreal but rarely supports open-source alternatives. These are not trivial barriers.

But what has changed is a trust crisis. In September 2023, Unity introduced a runtime fee based on game installations, posing an existential threat to free-to-play and hypercasual developers. While Unity eventually revised the fee, the damage was done. HSBC Global Investment Research reported that Godot doubled its share of new Steam releases from 1.3% to 2.7% between 2022 and mid-2024. Mega Crit, developer of Slay the Spire, switched from Unity to Godot after two years of development.

This is the same playbook that has repeated across enterprise software for thirty years. AT&T raised Unix licensing fees, and Linux emerged. Oracle’s pricing pushed enterprises toward PostgreSQL. As analysts at Jefferies put it, “pricing aggression by dominant vendors accelerated migration to open alternatives, which then matured from niche solutions into mission-critical infrastructure as the functionality gap closed.”

There is also a new structural reason why open-source is gaining ground specifically now, and it is directly tied to AI. AI coding tools work fundamentally better with open codebases. They can index the full source, generate contextual code, and understand the engine’s architecture in ways that closed-source engines simply do not allow. Cocos, another game engine, went fully open-source in January 2026. The stated rationale was explicitly about AI: open code is the fastest path to a fully AI-native development environment. Cocos 4 now ships new features as Model Context Protocols (MCPs) and AI agents rather than traditional libraries, a fundamentally different strategy designed for AI agents interactive with the engine programmatically.

Where the Ceiling Is (For Now)

open-source engines have real limitations today. Godot adoption remains concentrated in indie and lower-budget projects. Console SDK integration remains a barrier, though we expect console platform holders to gradually relax their proprietary SDK restrictions as open-source engines achieve broader commercial adoption.

Additionally, Unreal’s rendering technology (Nanite, Lumen, MetaHuman) has no open-source equivalent today, and its moat in high-fidelity AAA is real. But the “AI is eating software” dynamic applies here, too. If AI can increasingly write sophisticated rendering code, physics engines, and graphics pipelines, then even the deeply technical software underlying Unreal’s advantage becomes less defensible over time.

Where Value Accrues

The emerging architecture is a modular stack: AI tooling on top, open-source runtime on the bottom, and the traditional monolithic engine gets squeezed out. But we are skeptical of most AI tools themselves capturing durable value. Swapping one AI asset generator for another is trivial. There is no project lock-in, no accumulated data, and no switching cost. If a better model comes out next month, you just point your pipeline at it. The underlying foundation models are commoditizing fast, and most AI creation tools are relatively thin application layers on top of them. Unless there are network effects or proprietary data moats, AI productivity gains accrue to the end users and model providers, not the tools in between.

So if the engines are being unbundled and the AI tools are commoditizing, where does value actually land? We see three places:

1. Studios with strong IP and distribution. AI makes production cheaper and faster, which means that creative vision and audience become the moat, not technical capability.
2. The foundation model providers (OpenAI, Anthropic, Google) whose models power the AI tools.
3. Game development infrastructure where real switching costs exist through accumulated data and network effects.

Takeaways: The monolithic game engine is being pulled apart from both ends. AI-native tools are decoupling content creation, animation, and scripting from the engine, while open-source runtimes are commoditizing the underlying infrastructure. Unity is structurally exposed to this unbundling because its real business (advertising and monetization) depends on developers staying inside the engine ecosystem.

The historical precedent from enterprise software is clear: monolithic platforms get decomposed into modular, open, specialized layers once the functionality gap closes, and a trust-breaking event catalyzes migration. Gaming is following this pattern now. The open-source transition will not displace proprietary engines in AAA anytime soon, but for the growing majority of development happening at the indie, mobile, and mid-market level, the future looks like a composable stack of AI-native tools running on open-source.

‍