EA Sports puts 11,000 AI-generated college athletes on the field

How do you scan 11,000 real athletes into a video game without spending your entire production budget or driving your dev team insane? EA Sports just cracked the code with AI.

After an 11-year drought caused by likeness lawsuits, the publisher finally launched EA Sports College Football 25. The headline feature is an absurdly massive roster of 11,000 real NCAA Division I players, all built using generative AI.

Ditching the 3D scanner for headshots

When EA builds Madden NFL, putting players through full-body 3D photogrammetry scanning rigs is standard procedure. But college football has roughly six times the number of athletes as the NFL. Scanning everyone manually was simply a non-starter.

Instead, as first reported by The Wall Street Journal, EA built custom internal AI tools that convert standard 2D college headshots into detailed 3D digital avatars.

It was not entirely turnkey. The AI rarely nailed player likenesses on the first pass, so EA artists repeatedly adjusted the input data to coax out usable faces, simultaneously training the model on what quality standards to expect.

While the resulting models lack the ultra-fine pore details of Madden superstars, they are miles ahead of the generic, recycled preset faces from previous generations.

The $600 NIL playbook

EA killed the beloved college franchise in 2013 after losing a class-action lawsuit over unpaid student-athlete likeness rights. Following changes to NCAA policies, EA announced the reboot in 2021 with player compensation front and center.

With player opt-ins, each athlete received at least $600 and a deluxe copy of the game. High-profile stars negotiated larger payouts to appear on the cover and hype the title across social media.

Crucially, this pipeline sets up the foundation for future annual releases. With 11,000 models cataloged and the AI pipeline battle-tested, EA can churn out future editions far faster, with plans to export the tech to hockey and racing franchises.

Why it matters

We think EA played this brilliantly by understanding that sports fans care far more about authentic identity than pixel-perfect pore resolution.

For gamers, seeing their actual campus heroes on screen rather than generic randomized avatars completely transforms the immersion, even if the graphical fidelity sits slightly below Madden. It proves AI in game development does not have to be an all-or-nothing proposition where whole games are generated; using it as a targeted production accelerator delivers immediate player value.

Across the industry, this is a clear playbook for handling massive-scale asset generation. Studios working on sports titles, MMOs, or open-world games with tens of thousands of named NPCs will almost certainly adopt similar headshot-to-avatar pipelines.

That said, EA is not completely out of the woods. The flat $600 payout worked because players were eager to see the franchise return, but if College Football 25 turns into a cash cow, low-cost AI licensing could face fresh pushback from athletes demanding a bigger slice of the pie.