The underrated AI bull case for RBLX: self-improving games.
Roblox’s biggest competitor? It's not fortnite.
It’s tiktok, instagram reels, youtube. The average person spends 3 hours a day on these, and Roblox has one huge advantage over them: configurability.
Once a tiktok is posted, you can measure retention and engagement, but you can’t improve that exact same video for the next viewer. Everyone is stuck watching the original version.
But on Roblox, the learning from player engagement data can be fed back into the same game, incrementally improving it for all of the players that come after.
That loop exists today, but it is still slow. Creator teams have to look at analytics, form hypotheses, run A/B tests, ship changes, and analyse results manually.
The AI bull case is that this loop gets much faster.
AI detects where players are dropping off, proposes changes, rolls them out gradually, measures retention/replay/co-play/bookings, keeps what works, and repeats.
That matters more than “AI can generate 3D assets” or “AI can make a whole game.” Those are useful, but I don’t think they are the killer use case.
The killer use case is Roblox turning its existing infrastructure and engagement data into a platform where millions of live experiences continuously improve.
Tiktok can optimise the feed. Roblox can optimise the actual entertainment. I think the long term opportunity for attention is much bigger than people are pricing in.