Epic Games is going AI-first. The first thing the pivot delivered is a delay.

Bloomberg reported on October 12 that Epic Games has delayed its collaboration with Disney, in part because the project will now be built under an "AI first" model, with systems such as Claude handling the code and the art. The plan had already circulated inside the company as an email. Within hours the chief executive, Tim Sweeney, confirmed the pivot in public, and his post is the only part of the story with a named author. Two days later, the newest artefact the strategy has produced is a delay.

What the report says the pipeline will do

The Bloomberg story, by Jason Schreier, rests on unnamed people familiar with the project and on internal employee messages the outlet says it viewed; no source is identified and no document is published. According to Ars Technica's account, human developers will be moved to managing and guiding code and art assets produced by models including Anthropic's Claude. That is where the report's two most quoted fragments come from: no more hand-written code, and assets generated with full AI workflows.

Epic's own public version is narrower than the reporting. A spokesperson said the company is "embracing AI-accelerated development, led by Epic developers who are the visionaries and creative drivers for everything we do". That statement does not name Claude, does not mention code, and does not mention the Disney project. The company has published no technical description of the workflow, and it did not put out the internal email.

A $1.5 billion shooter with no date

The Disney deal was announced in early 2024, alongside a $1.5 billion investment by Disney into Epic, and the first product was to be a Marvel- and Star Wars-licensed shooter slated for March 2027. It is now delayed indefinitely, with no new date communicated, and the same reporting attributes part of the delay to development that has been understaffed and slower than expected.

For Sweeney this is a direction he has argued for years rather than a reversal. He told IGN in June that AI prompts would soon let a ten-person team build something the size of a major open-world game, and last November he said Steam's AI-use disclosure tags make no sense because AI tools will be involved in nearly all future production. The new element is not the thesis. It is that a named, licensed, billion-dollar project is being handed to it.

"Not the same stuff with fewer people"

Sweeney posted on X on October 12, shortly after the Bloomberg story went up. His framing is entirely about capacity: "The aim isn't to do the same stuff with fewer people, but to enable everyone here to build MORE and BETTER stuff, and to bring the benefits to all teams using Unreal." Replying to a critic who said he could not "read the room", he wrote that much of the room wants it to be 2016 again, that it is not, and that Epic's job as a supplier of tools to the industry is to design for the present and the near future, a reply the Forbes write-up quotes in full.

Read them as a pair and the argument is about intent, not scale. Nothing in either post says how many people will work on the Disney title, what it will cost, when it will ship, or what "BETTER" is being measured against. Sweeney is telling his audience what he wants the pivot to mean; the claim is his, and it is unfalsifiable until something ships.

The claim that has no headcount

"More and better" and "cheaper" are not mutually exclusive readings of the same plan, because Epic has already taken the cheaper route and documented why. In March 2026 the company laid off more than 1,000 employees, citing a downturn in Fortnite engagement, and raised V-Buck prices the same month to help pay the bills, with Fortnite in the red. A pipeline sold on doing more with the same people is, in that context, also a pipeline that could do the same with fewer.

The internal version of the promise has been circulating longer than the public one. Stephanie Arnette, a senior external development manager on Fortnite, has said the company knows the fear is that AI takes jobs, that it is not the goal, and that the goal is greater efficiency. The only shipped evidence the public has about Epic's AI pipeline points the other way: in May 2025 an AI-voiced Darth Vader added to Fortnite started swearing and voicing odd opinions until a hotfix removed it. One anecdote is not a measurement, but it is the sole artefact anyone outside Epic can look at.

Studios that never announce it will follow anyway

Paul Tassi's Forbes analysis makes the case that the announcement matters less than the industry's direction. Other studios will pursue the same tools whether or not they say so, most AI use will be invisible to buyers, and consumers can vote only where AI content is visible — which is why he notes that Google already has the majority of its code written by AI. The comparison to hold against that is what the industry's tool use actually looks like: a survey of more than 2,300 game professionals ahead of GDC 2026 found 36 percent using generative AI at work, 30 percent at development studios, and among those users 81 percent using it for research and brainstorming, 47 percent for code assistance, 35 percent for prototyping.

That distribution is a long way from a studio whose code is entirely model-written. AI has already been shown doing one piece of game work end to end — coding agents rebuilt most of Modern Warfare 2's source — but that task had a machine-checkable answer and no creative judgement in it, which is precisely the property a shipped game does not have. The loudest outside voice on the news was not technical: Nikita Bier, Twitter's former head of product, called the timing beautiful, with the most expensive game ever made arriving the same month AI could supposedly produce comparable experiences, and named it the swan song for the big-studio era. He was referring to GTA 6. The word carrying that claim is "comparable", and the replies said so.

What would settle it

There is a temptation to treat this as a verdict on AI in game development, and the material does not support one. Epic has published no shipped game from the new pipeline, no productivity figure, no quality comparison, and no budget line. The claim that an AI-first pipeline produces a better game than a conventional one is, on the public record, a prediction by the person selling the tools.

The stakes are asymmetric, and that is the part that will not be resolved by argument. If the pipeline works, other studios adopt it quietly and the labour question is never put to them in public. If it fails, the cost lands on the project and on the people attached to it, not on the post announcing it. What to watch is narrow: whether Epic ever attaches a date to the Disney title, whether it publishes any number of any kind, and whether the next studio to make this move says it out loud. Sweeney's post drew roughly 15,000 likes by way of public support, and the most-liked reply, from a developer telling him he could not read the room, drew considerably more. The disagreement is live, and neither side has a shipped artifact to settle it with.