Key Takeaways:
- Masahiro Sakurai, creator of Super Smash Bros. and Kirby, warns that AI will massively increase the number of game releases, but most will be unoriginal and unprofitable.
- Speaking to Denfaminicogamer, he noted Steam went from 275 games in 2010 to over 19,000 in 2026 — and expects that figure to climb further due to AI.
- He predicts AI-generated games will mostly reproduce systems already seen in other titles rather than creating anything genuinely new.
- Sakurai pointed to Ico and Shadow of the Colossus creator Fumito Ueda as an example of human creativity that AI simply cannot replicate.
- While AI can analyse and copy existing works, it cannot create something that doesn’t yet exist — and that’s where human vision matters most.
- Steam Next Fest 2026 already featured more than double the demo count of the previous year, partly attributed to an influx of AI-generated titles.
- Fumito Ueda himself has stated Gen Atlas — his upcoming game — was developed without generative AI for core game creation.
Masahiro Sakurai doesn’t do many interviews. The creator of Kirby and Super Smash Bros. is famously private about his current projects and thoughtful about when he speaks publicly. So when he sits down with a major gaming outlet and talks about the state of the industry, it’s worth paying close attention. In a recent interview with Japanese outlet Denfaminicogamer, Sakurai addressed something every developer, player, and publisher is thinking about right now: what happens to games when AI can make more of them, faster, and cheaper than ever before?
His answer is genuinely worth sitting with — both the concern it raises and the reason for hope buried inside it.
19,000 Games a Year on Steam — and That’s Just the Beginning
Sakurai set the scene with a number that really does take a moment to land. When he began judging for the Game Designers Awards in 2010, approximately 275 games were published on Steam in a year. In 2026, that figure has surpassed 19,000. The awards, which he recently retired from judging, have watched the industry transform from a manageable catalogue into something that defies any individual’s ability to keep up with.
“19,000 is amazing, but I think that number will increase even more from now on,” Sakurai said in the interview (translated from Japanese). “I’m sure of it. That’s because AI will increase the number. However, I think most of them will become ‘unprofitable titles’. There are too many to keep track of, and it’s impossible to understand them all.”
The math is already playing out. Steam Next Fest 2026 featured more than double the number of demos compared to the previous year, with a significant portion of that increase attributed to AI-assisted game production. Developers who have spent years building original projects have started pulling planned demos and playtests out of concern that their work could be copied by AI tools before it properly launches — a chilling new competitive pressure that didn’t exist a few years ago.
The AI Slop Problem — Reproductions of Reproductions
Sakurai’s specific concern isn’t just volume — it’s quality and originality. Or rather, the lack of both. When it comes to AI-generated games, he argued, the output will predominantly be games that reproduce systems and concepts already seen elsewhere, rather than creating anything new.
“When it comes to things created with AI… or things mass-produced with poor quality, I think most of them will simply be reproductions of systems seen in various other games,” he said. “I think there are already a lot of those now.”
He used a concrete and somewhat funny example to illustrate the point: games like Librarian: Tidy Up the Arcane Library! — which challenges players to organise books on shelves — and Needle in a Haystack Simulator, which went viral on the strength of asking players to find a single needle among five million pieces of hay. Simple, clever, niche. But the moment Needle in a Haystack Simulator went viral, AI-assisted copies began appearing on Steam almost immediately, prompting the original developer to add a bright red “original” badge to their Steam artwork just so players could tell which version was the real thing.
That’s the dynamic Sakurai is describing. AI doesn’t need to understand why something works to copy the structure of it. It just needs the thing to exist.
But Here’s the Thing — AI Can Only Reference What Already Exists
This is where Sakurai’s argument gets genuinely interesting rather than just cautionary. He’s not saying AI is capable of doing what the best human designers do — he’s saying it is fundamentally incapable of it by nature of what AI actually is.
“Basically, AI refers to things that have been created so far, but it can’t refer to things that haven’t been created before,” he explained. “I think this is true not only for AI, but for various kinds of creative works as well.”
To illustrate what he means by truly unreplicable human creativity, Sakurai pointed to Fumito Ueda — the game designer behind Ico, Shadow of the Colossus, and The Last Guardian, three of the most singular artistic statements in the history of the medium. Sakurai acknowledged that a capable AI could potentially analyse Ueda’s existing games and produce something that resembles their atmosphere or structure. But that’s not the same thing as what Ueda will create next.
“But even if AI can create all these sorts of works, such as by analyzing the atmosphere of games from Fumito Ueda like Ico and Shadow of the Colossus and using that to create something similar, which could happen — AI would not be able to create whatever Ueda himself will create in the future. What kind of artistic style will his future works possess? That’s something only he can create. I think that’s where human creativity lies.”
The point is precise and worth unpacking. AI is trained on what has been made. It can extrapolate, combine, and reproduce. What it cannot do is originate — it cannot create a concept that exists nowhere in its training data, because it has no source for it. A Fumito Ueda game is not a recombination of things that already exist. It comes from a specific human perspective, a set of obsessions and experiences and aesthetic instincts that belong to one person and can’t be modelled from the outside.
Ueda’s Response Is Already Embodied in His Next Game
Interestingly, Fumito Ueda himself — the creative whose future work Sakurai uses as the example of unreplicable human artistry — has already taken a clear position on generative AI in his upcoming game. Gen Atlas (formerly known as Project Robot), developed by Ueda’s studio genDESIGN and published by Epic Games, was confirmed to have been built without generative AI for core game creation. In a games industry where more and more studios are quietly integrating AI tools into pipelines, Ueda drew a deliberate line.
For fans of Ico and Shadow of the Colossus, knowing that Gen Atlas was made the same way those games were — by hand, by people, with a specific human vision driving every decision — adds something meaningful to the anticipation around it. It also makes Sakurai’s words land differently: he’s not speaking theoretically about some abstract future creator. He’s talking about a designer who is actively making a game right now, under the same conditions he’s describing.
What Sakurai Is Actually Saying About the Future
There’s a version of Sakurai’s argument that reads as simple pessimism about AI. But that’s not quite what he’s doing. He’s drawing a distinction between two categories of creative work — work that remixes and reproduces what already exists, and work that creates something genuinely new — and arguing that only the latter has lasting value, and only humans can do it.
“People who forge their own path… people who understand and can create their own unique flavour… will become increasingly important from now on,” he said. “I think this is especially true in an era of mass production and low-quality goods.”
In other words: as the noise level rises, the signal gets more valuable. If 19,000 games a year becomes 50,000 games a year — most of them AI-assisted reproductions of existing ideas — then the games that offer something genuinely original become harder to find but more precious when you do. The challenge is discoverability in a flooded market, which is a real and serious problem. But the solution isn’t to match the flood with more of the same. It’s to be the thing that the flood can’t replicate.
Sakurai didn’t say anything about what he himself is currently working on — that remains his own closely guarded creative secret. But in speaking about Ueda’s future work as something AI fundamentally cannot produce, he’s implicitly making a claim about his own as well.
Some things can only come from one person. That’s not a limitation of AI. It’s the whole point of having Masahiro Sakurai, or Fumito Ueda, or anyone whose creative identity is genuinely their own.
For more on the gaming industry this week, read our coverage of Zenith: Nexus shutting down in December — a story about what happens when a live service game can’t sustain itself in an increasingly crowded market. And if you want something more upbeat, Call of Duty: Modern Warfare 4 is live now with Shipment dropping on October 27.
