Google Built a Roblox Killer—Then Locked Out the Kids

Google has spent years trying to figure out how to win in gaming.

It tried selling games through Google Play.

It tried game streaming with Stadia.

And Stadia ended exactly the way many Google experiments do: the technology worked, but the business never found enough traction.

When Google announced Stadia's shutdown in September 2022, it said the service had not gained the user adoption it expected. Google subsequently refunded Stadia hardware purchases and game and add-on purchases, while the servers finally went offline on January 18, 2023.

Now Google is back.

But this time it isn't trying to convince consumers to play Google's games.

It is giving consumers the tools to make the games themselves.

That difference may be enormous.

Google has reversed the Stadia bet

Stadia's fundamental assumption was that the future of gaming was about distribution.

You don't need an expensive console.

You don't need a powerful gaming PC.

Google's data centres do the hard work. You simply stream the game to whatever screen you own.

It was technically elegant.

But gaming is not Netflix.

People don't just consume games.

They build identities around them. They form communities. They create mods, maps, characters, videos and entire ecosystems around games.

Stadia tried to remove the hardware barrier.

It did not remove the content and community problem.

Playground is approaching the problem from almost the opposite direction.

Google's new experimental platform allows users to describe a game in natural language and have AI generate a playable browser game. Users can then continue modifying gameplay, graphics and rules through conversational instructions.

You don't have to learn a programming language first.

You don't have to understand game engines.

You don't necessarily have to know how physics systems work.

You can simply say what you want.

The computer becomes the game developer.

And that is a much more interesting bet.

Google is attacking the creation bottleneck

For decades, making a game required a stack of specialised skills.

Programming.

Game design.

3D modelling.

Animation.

Sound.

Level design.

Physics.

User interfaces.

Testing.

A teenager with an idea could have the imagination to invent the next great game and still be completely unable to build it.

AI changes that equation.

Google Playground essentially says:

Describe the game.

Then:

Change it by talking to it.

Want the character to jump twice as high?

Ask.

Want the floor to become lava?

Ask.

Want enemies to move faster?

Ask.

Want a completely different visual style?

Ask.

Google's own documentation says games can be modified through conversational instructions, including gameplay logic, graphics and rules.

That is not merely an easier game engine.

It potentially changes who gets to be a game developer.

And Roblox already proved the demand

This is where Roblox becomes extremely important.

Roblox didn't become a gaming giant because it produced the world's best professionally developed games.

It built an ecosystem where users could create and distribute experiences to other users.

The result is extraordinary scale.

Roblox reported $4.9 billion in revenue in 2025, up 36% year over year. It also reported an average 127 million daily active users across more than 180 countries, with users spending 123.9 billion hours on the platform during the year.

The important number isn't just the revenue.

It is the amount of human creativity happening inside the platform.

Roblox says its community creates the experiences that attract users, with millions of experiences available on the platform.

That is the real Roblox machine:

Creators make things → players arrive → players become creators → more content attracts more players → the ecosystem gets bigger.

Google isn't inventing this model.

It is trying to remove the programming barrier from it.

And that could be dangerous for Roblox

Roblox Studio still requires creators to understand a development environment and, depending on what they are building, scripting and other technical concepts.

Google's pitch is radically simpler.

Talk to the machine.

That matters because the biggest potential population of game creators is not professional developers.

It is people who have ideas but cannot code.

Millions of them.

A child who has never written a line of Lua could imagine a game.

A teacher could build an educational game.

A TikTok creator could turn a viral meme into an interactive experience.

A musician could create a game around a song.

A football fan could build a football challenge.

A teenager could spend an afternoon experimenting with increasingly ridiculous ideas.

The barrier becomes imagination rather than programming.

And whenever technology moves a capability from specialists to ordinary people, the amount of content can explode.

But then Google made a strange decision

Playground is 18+ at launch.

Google's official help centre says the platform is currently available in the United States to users aged 18 and over, with additional countries planned.

That is where the strategy becomes fascinating.

Because Roblox's audience is heavily populated by younger users.

In Roblox's latest age-checked data, among users who had completed an age check, 35% were under 13 and another 38% were 13–17. Only 27% were 18 or older. Roblox cautions that these proportions can change as more users complete age checks, so they should not be treated as the composition of its entire user base.

But even with that caveat, the implication is obvious.

A huge part of the audience that made Roblox's user-generated ecosystem so powerful is below Google's current age threshold.

Google has built an AI system designed to eliminate the technical barrier to game creation.

Then it has restricted the product from the demographic that has spent years proving that amateur game creation can become a massive business.

That isn't necessarily a mistake.

It may be the most revealing part of the launch.

Because AI changes the child-safety equation

Giving a child Roblox Studio is not the same thing as giving a child a generative AI system capable of producing software.

A young creator using conventional tools still has to learn something.

AI removes much of that friction.

That creates extraordinary creative possibilities.

It also creates extraordinary moderation problems.

A sufficiently capable generative system can potentially be prompted to create violent, sexual, hateful, exploitative or otherwise inappropriate material.

Google's Playground rules explicitly prohibit sexually explicit material, exploitation of children, regulated goods, harmful activities and other categories of abuse. Published games also undergo automated safety screening before being publicly listed.

And Google already has broader generative-AI policies addressing dangerous activity, malware, privacy violations and attempts to circumvent safety systems.

The problem is that game creation is generative AI with an interactive output.

The AI isn't merely producing an image.

It is constructing something people can play.

The user can then iterate.

And iterate again.

And share it with other people.

That creates a moderation loop far more complicated than simply filtering a chatbot response.

Imagine the 12-year-old problem

This is where the business opportunity collides with liability.

Imagine a 12-year-old telling an AI:

“Build me a game where…”

The model creates something.

The child changes it.

The game becomes popular.

Thousands of other children play it.

Someone discovers that the game contains inappropriate material.

Someone's likeness has been copied.

A copyrighted character has been reproduced.

A dangerous challenge is embedded in the mechanics.

A stranger uses the game's social features to contact minors.

Now the question becomes:

Who is responsible?

The child?

The parent?

Google?

The AI model?

The platform?

The person who published it?

The company that provided the tools?

That is not a technical question.

It is a legal and political nightmare.

And Google knows it.

Roblox has already been forced into this fight

Roblox itself has been moving aggressively toward age assurance.

In 2025, it announced plans to expand age estimation for users accessing communication features, using facial age estimation, ID verification and verified parental consent.

By February 2026, Roblox said 45% of its 144 million daily active users had completed an age check; among age-checked users, 35% were under 13, 38% were 13–17 and 27% were adults.

That tells you something important.

Even the company that built one of the world's largest user-generated gaming ecosystems is having to redesign its platform around age verification and age-appropriate interaction.

Google is entering that environment with something even more powerful:

generative AI capable of creating the content itself.

The safety burden is therefore substantially higher.

Google's real problem isn't creating games

Google can create games.

That is no longer the interesting question.

The difficult question is:

Can Google create billions of games safely?

That is completely different.

Generating one game is a demonstration.

Generating one million games is a platform.

Generating one billion games while simultaneously determining:

• what is safe,
• what is illegal,
• what infringes copyright,
• what exploits children,
• what constitutes harmful behaviour,
• what violates trademarks,
• what contains malware,
• what manipulates users,
• what should be promoted,
• what should be removed,

is a completely different engineering problem.

And then comes the even harder question:

Can you do all of that in real time?

This is why the token economy matters

Playground is free to play.

But game creation uses tokens.

Google says the free tier includes a limited number of weekly generation tokens, while Google One AI subscribers receive higher weekly limits.

This is a clever business model.

Google doesn't need to charge every person to play.

It can let the audience consume games for free while charging the creators for the computational resources required to generate them.

That creates a familiar platform dynamic:

Players generate attention. Creators generate content. AI generation consumes compute. Google monetises the compute.

And suddenly Google One becomes more than cloud storage and AI subscriptions.

It becomes a potential creative-compute subscription.

The more you create, the more tokens you consume.

The better the model becomes, the more expensive large-scale creation could become.

Google therefore has an incentive to make creation addictive.

But Unity makes the story even bigger

Google is not doing this alone.

It has partnered with Unity on Unity Spark, a more advanced AI-assisted game-development product expected to enter closed beta later this year.

The idea is to move beyond simple browser experiments toward higher-fidelity mechanics and professional game development. Games made with Unity Spark are expected to be playable through Playground.

That creates a potentially powerful pipeline:

Google AI → Playground → Unity → professional game creation.

If successful, Google isn't merely building a toy for people who want to make silly browser games.

It could be building an AI interface to one of the world's major game-development ecosystems.

That is much more strategically significant.

The bigger threat to gaming may not be AI-generated games

There is an even deeper possibility.

AI could change the economics of game supply.

Today, games are expensive to produce because human labour is expensive.

A professional game can require hundreds or thousands of people.

If AI dramatically reduces the cost of producing functional games, the world could suddenly experience an explosion of interactive content.

The bottleneck moves.

We go from:

“Who can afford to make a game?”

to:

“Who can get anyone to play it?”

That sounds familiar.

It is exactly what happened to websites, blogs, YouTube videos, podcasts and social media posts.

Once creation becomes cheap, distribution and attention become the scarce resources.

That means Google Playground isn't necessarily just competing with Roblox.

It is entering the same attention economy as YouTube, TikTok, Roblox, Discord and social gaming.

And Google has one enormous advantage

Google understands discovery.

Search.

YouTube.

Android.

Chrome.

Google Play.

Ads.

Cloud.

AI.

Few companies have as many pieces of the digital distribution puzzle.

Imagine a future in which someone creates a game with Playground, publishes it, and that game can be discovered through Google's ecosystem.

Google could potentially connect:

AI creation + gaming + search + video + social discovery + advertising + cloud infrastructure.

That is a much more formidable proposition than Stadia ever was.

Stadia primarily wanted to control how games were delivered.

Playground could help Google control how games are created and discovered.

That is a fundamentally different position.

But Google has been here before

This is where the Stadia comparison becomes useful.

Google has a long history of entering markets with extraordinary technology and discovering that technology alone doesn't create an ecosystem.

Stadia demonstrated this brutally.

Google itself admitted that its streaming technology had a strong technical foundation but that Stadia had not achieved the user traction it expected.

The lesson is important:

Technology is not the same thing as a market.

Google can build an impressive game-generation model.

But users still need reasons to return.

Creators need incentives.

Players need communities.

Popular games need discovery.

Creators need ways to earn money.

Parents need confidence.

Developers need intellectual-property protection.

And Google needs to convince everyone that Playground will still exist five years from now.

That last point matters because Google has a reputation for killing products that don't reach sufficient scale.

A creator deciding whether to invest hundreds of hours building a community does not just ask:

“Is this tool good?”

They also ask:

“Will Google still be here?”

Google hasn't actually built a Roblox killer yet

This distinction matters.

Calling Playground a “Roblox killer” is premature.

Playground is currently an experimental platform, launched in the U.S. for adults, with game creation rolling out gradually.

Roblox has a decade-plus ecosystem, enormous social graphs, established creator economies, millions of experiences and more than 127 million average daily active users in 2025.

Google has a fascinating technology.

Roblox has an enormous network.

And networks are difficult to kill.

The interesting question is whether AI changes the equation enough.

The real Roblox moat may disappear

Roblox's historical advantage was not simply that people could create games.

It was that Roblox made creation accessible enough for ordinary users.

But AI could reduce the creation barrier even further.

If a child once needed to learn scripting to become a creator, and tomorrow someone can simply describe a game to an AI, then the definition of a “creator” changes.

The next generation may not think of themselves as programmers.

They may think of themselves as directors.

They describe worlds.

The AI constructs them.

They iterate.

They publish.

They build communities.

That is potentially much bigger than Roblox.

But it creates another question:

What happens to the value of the human creator?

The internet is about to have a content problem

We already have too much text.

Too many images.

Too many videos.

AI is now making all three dramatically cheaper.

Gaming may be next.

Imagine millions of playable games being generated every day.

The supply could become infinite.

Then discovery becomes almost impossible.

Nobody can play a million games.

Nobody can watch a million videos.

Nobody can read a million articles.

AI therefore creates a paradox:

It makes creation abundant and attention scarce.

The winners may not be the platforms that generate the most games.

They may be the platforms that are best at determining which game you should play next.

And that puts Google in an interesting position.

Because Google has spent decades building recommendation and discovery systems.

The biggest prize really is the young creator

The user's original observation about children gets to the heart of the matter.

The most important future gaming population isn't necessarily today's professional developer.

It is the teenager who currently spends three hours playing a game and could tomorrow spend those three hours building one.

Roblox understood that.

Its ecosystem turns players into creators.

Google understands it too.

But generative AI introduces a new level of risk.

Google can currently avoid much of that risk by saying:

18 and over.

That is commercially frustrating—but strategically understandable.

The adult market is smaller than Roblox's youth-heavy ecosystem.

But it is much easier to manage.

Google can experiment.

It can measure abuse.

It can improve moderation.

It can develop age assurance.

It can learn what happens when millions of AI-generated games begin circulating.

And perhaps eventually it will try to solve the harder problem:

How do you safely give children the same creative power?

Whoever solves that problem has something enormous

The ultimate prize isn't a better game engine.

It isn't even a better AI model.

It is a platform where anyone can imagine a game and immediately create it, while the platform can safely distribute that content to millions of people.

That would collapse one of the biggest barriers in entertainment.

A 13-year-old with an idea could potentially compete with a studio employing hundreds of developers.

A small African developer could create interactive experiences without needing a massive technical team.

An educator could create a simulation in an afternoon.

A filmmaker could turn a story into an interactive world.

A musician could build a game around a song.

A teenager could invent something nobody at a major studio thought of.

That is the extraordinary promise.

But the same technology can also generate harmful material at unprecedented speed.

So the defining problem of AI gaming may not be creation.

It may be governance.

Google's Playground is therefore an admission—and a warning

Google learned from Stadia that owning the infrastructure isn't enough.

Now it is betting on something much closer to the heart of gaming:

people creating things for other people.

That is why Playground is potentially more important than Stadia ever was.

Stadia tried to make the console disappear.

Playground is trying to make the game developer disappear as a technical bottleneck.

And the irony is extraordinary.

Google has built a system that could eventually make game creation almost effortless.

Yet it has launched that system by excluding the demographic that has arguably demonstrated the strongest appetite for user-generated gaming.

That isn't necessarily evidence that Google doesn't understand the market.

It may be evidence that Google understands the risk better than the opportunity.

The next Roblox will not be won by whoever generates the prettiest AI game.

It will be won by whoever solves the impossible-looking equation:

children + generative AI + creation + social interaction + monetisation + safety.

Roblox has the children.

Google has the AI.

Unity has the development infrastructure.

The company that figures out how to put all of those pieces together safely may own the next generation of gaming.

And that is why Google's most important gaming product in years isn't really about playing games.

It is about giving everyone the ability to make them.