Can Epic Games survive the generative AI revolution — or will the engine that built a thousand virtual worlds become another legacy software empire?
Salve, users.
On October 8, Epic Games promoted Unreal MCP, its implementation of the Model Context Protocol (MCP), an open standard allowing AI agents to communicate directly with software.
Unreal Engine 5.8 already includes an experimental MCP plugin, while Unreal Editor for Fortnite (UEFN) has its own integration. AI agents such as Claude Code, Codex and Cursor can now interact with projects, create objects, configure lighting, write code and run tests.
In other words, instead of spending six hours learning how to position a virtual lamppost, you might simply ask an AI to build an entire street.
For the metaverse, this could be revolutionary.
The Adobe problem: when software becomes the obstacle
Remember when Adobe was indispensable?
Creative Cloud remains enormously capable, but subscription fatigue, intimidating complexity and increasingly accessible alternatives have eroded its aura of invincibility.
Unreal faces a related existential challenge.
Its pricing is actually more creator-friendly than Adobe’s: game developers generally pay royalties only after $1 million in product revenue, while qualifying smaller businesses use the engine free. Larger non-game studios can face $1,850 annual seat licences.
The bigger problem isn’t price. It’s friction.
Why learn an elaborate technical ecosystem when generative AI promises to create environments through natural language?
Google’s new Playground and forthcoming Unity Spark are pushing conversational game creation. World Labs’ Marble generates persistent, navigable 3D environments. Google’s Genie explores real-time, AI-generated worlds.
Suddenly, mastering the intricacies of traditional game development looks less like a superpower and more like learning medieval Latin.
Epic understands the threat.
Its response is not to fight AI, but to absorb it.
Open AI, closed worlds?
Here’s where things get interesting.
MCP is model-agnostic. Creators can potentially choose competing AI assistants rather than becoming dependent on Epic’s preferred intelligence provider.
That’s excellent.
But model-agnostic does not mean platform-agnostic.
MCP makes Unreal easier to operate. It does not automatically make Unreal worlds compatible with Unity, Godot, browser-based environments or independent metaverse platforms.
This distinction matters enormously.
Epic’s Unreal Engine 6 roadmap promises greater interoperability using standards such as glTF and Universal Scene Description (USD), alongside portable content and cross-game economies.
These ambitions extend far beyond MCP.
A genuinely open metaverse requires transferable assets, identities, behaviours, economic relationships and experiences. Not merely an AI assistant capable of rearranging furniture inside someone else’s proprietary kingdom.
Otherwise, we’ve liberated the builders while leaving the landlord in charge.
The death of the Unreal aesthetic?
There’s another problem.
For all its astonishing graphical power, Unreal has encouraged a recognisable visual vocabulary: photorealistic environments, cinematic lighting, familiar asset libraries and endless variations of dystopian urbanism.
A world of limitless imagination, apparently designed by the same architectural intern.
To be fair, this isn’t an intrinsic engine limitation. It’s frequently a failure of artistic direction.
MCP could help break that cycle.
Creators might orchestrate different generative models, procedural tools, custom materials and unconventional visual workflows without mastering every technical detail themselves.
Epic is also developing diffusion-based workflows that use scene geometry, camera information and depth maps to guide stylised outputs.
Imagine worlds inspired by Byzantine mosaics, Japanese woodcuts, African futurism or surrealist cinema — rather than yet another demo resembling a luxury car commercial after the apocalypse.
But there’s a catch.
Generative AI can democratise aesthetics or industrialise mediocrity. If everyone uses identical models, prompts and presets, we’ll simply replace Unreal slop with AI slop.
The technology cannot substitute for imagination.
Who builds the next metaverse?
Unreal isn’t dying tomorrow.
Its advantages in physics, rendering, interactive systems and production pipelines remain substantial. Generating a beautiful virtual city is one thing. Making it persist, function and support thousands of interacting citizens is quite another.
MCP could transform Unreal from intimidating specialist software into an accessible creation infrastructure.
But the real prize isn’t preserving Epic’s dominance.
It’s creating a global, interoperable digital civilisation where anyone can build, experiment and participate without asking permission from a corporate gatekeeper.
The future of the metaverse shouldn’t belong to the company with the best engine. It should belong to the creators with the best imagination.
Epic has opened a door.
Now let’s see whether it opens the walls.
Ride or Die in the Metaverse.
Hiro Nakamoto ??
