Google Playground makes games from a prompt. Players don't.

On 7 October 2026 Google Labs launched Playground, a browser game maker where you type an idea and get a playable game back. It's the biggest name yet in a year of prompt-to-game tools, and it makes the cheap part cheaper. Getting played keeps getting harder. Moloco counted 190,000 mobile games released in 2025, and only 2,500 passed 500,000 downloads in their first year. The breakouts it studied ran about 2.5 times as many ad creatives as other launches. Players don't arrive with the game. They have to be found, and on mobile most of them are paid for.

What Google Playground is, and what it isn't yet#

Playground lives at playground.google. You describe a game in a chat window, play it straight away, then change it by asking: different physics, new scoring rules, another character, a whole new level. You can start from a blank canvas, remix a starter prompt or take guided help (Google). Because it runs in the browser, you play on a phone or a laptop.

The details from launch day:

• Who can use it: "users in the U.S. (18+)", says Google. TechCrunch reports more countries are planned.
• What it costs: browsing and playing are free. Creating uses a weekly token allowance, with higher limits on a paid Google AI subscription (TechCrunch).
• What's under it: Gemini, Nano Banana for images and Lyria for music, in a custom game harness (9to5Google).
• Your own art: TechCrunch reports you can upload visuals and the AI turns them into game assets in the game's style.
• Sharing: keep a game private, send a link, or publish it to the Explore gallery. Every published game is safety-screened.
• What's next: an integration with Unity Spark, which Google says brings "professional-level mechanics", high-fidelity 3D and the Unity runtime. Spark is in testing, with a closed beta "coming soon".

The output is simple for now. VGC calls them "simple AI games", and Inven Global describes them as relatively simple. Google hasn't said whether creators can earn money, export a game or ship it to the Play Store. For now, a Playground game lives on Playground.

If you want the wider picture of prompt-to-game tools and why their output is easier to change than most ad output, our earlier post on games generated from a prompt, and why their ads still take a week covers Higgsfield, Genie 3 and Roblox Cube.

Every prompt-to-game tool ends at the same question#

Making the game is the demo. Who plays it is the product question, and each company has answered it the same way: with its own discovery system.

Google's answer is the Explore gallery: "Our gallery uses player ratings and play activity to spotlight games," the launch post says. GamesBeat adds that Google will highlight standout creations across its network. In July Roblox launched Build, a mobile-first tab in the Roblox app that turns text prompts into a basic game. Build games join the same pool as every other Roblox game and face the same retention-based ranking. Roblox's own summary is blunt: "If no one plays it—no one can find it" (Roblox). By September, about 9,000 games had been published with Build, and 71% of Build users had never opened Roblox Studio (Roblox).

Prompt-to-game tools come with a gallery that ranks by engagement. Outside one, reach is something you buy and earn with creative.

	Google Playground	Roblox Build	Your game on the app stores
How games are made	Chat prompts in the browser	Text prompts in the Roblox app, on Studio's back end	Your team, your engine, your art
Where players find it	A link, or the Explore gallery	Roblox's discovery, in the same pool as every game	Store search, featuring, and paid ads
What decides who sees it	Player ratings and play activity	Retention, over the first 28 days and beyond	Your bids, your creatives and your store page
What you control	The game and who you share the link with	The game and its early retention	The ads, the targeting and the store listing

A gallery ranked by play activity is a loop: games that get played get shown, and games that get shown get played. A new game has to get into that loop somehow. Inside Playground, the way in is sharing a link and hoping. Outside it, the way in is paid acquisition, and paid acquisition runs on ads.

Supply outran attention before Playground arrived#

Prompt-made games are joining a market that was already crowded. Steam had more than 19,000 releases in 2025, according to SteamDB data reported by KitGuru. Sensor Tower counted another 12,000 Steam releases in the first half of 2026 alone, up 19% year on year (Sensor Tower). Then, in the last week of August, 720 games came out in seven days, the most ICO Partners has ever tracked.

One record week on Steam: 720 releases, by review countReleases from 24 to 30 August 2026, grouped by reviews at the time of counting

53095

• 0–9 reviews 530
• 10–49 reviews 95
• 50–99 reviews 33
• 100–999 reviews 52
• 1,000+ reviews 10

One record week on Steam: 720 releases, by review count

0–9 reviews	530
10–49 reviews	95
50–99 reviews	33
100–999 reviews	52
1,000+ reviews	10

Nearly three in four of that week's games had fewer than 10 reviews. And new games compete with old ones for the same hours: Valve's Steam Replay 2025 found that only 14% of all playtime that year went to games released that year (GamingOnLinux).

Mobile is bigger and just as steep. 42matters lists more than 316,000 games on Google Play and about 259,000 on the App Store today (42matters). Of the 190,000 mobile games released in 2025, 2,500 crossed 500,000 downloads in their first year, according to a Moloco report covered by PocketGamer.biz. That's about 1.3%.

Industry analyst Joost van Dreunen put the AI part of this plainly on the Mobile Dev Memo podcast, talking about discovery: "AI at least in its current manifestation just adds to the problem" (MDM).

On mobile, most players are bought#

Outside a curated gallery, the main way a mobile game reaches players is paid user acquisition. AppsFlyer's 2026 gaming report counts 24.8 billion game installs in 2025, and 14.1 billion of them were paid (AppsFlyer).

Share of mobile game installs that were paid, 2025

Hypercasual, Android83%

Hypercasual, iOS72%

All games, Android59%

Casual, Android54%

All games, iOS44%

Midcore, iOS24%

Share of mobile game installs that were paid, 2025

Hypercasual, Android	83%
Hypercasual, iOS	72%
All games, Android	59%
Casual, Android	54%
All games, iOS	44%
Midcore, iOS	24%

Buying those installs is getting more expensive. Adjust's 2026 gaming report, as summarised by GameDevReports, puts the global gaming cost per install at $0.56, up 30% in a year. In North America it went from $1.28 to $1.68.

Spending is up too: $25 billion went into gaming UA in 2025 (AppsFlyer). And the queue of buyers is longer. In the first half of 2026 the average month had 109,500 mobile game advertisers, a 31.1% rise on a year earlier, by SocialPeta's count (Game World Observer).

So the market has more games, more advertisers and pricier installs. A tool that makes games faster adds to the supply. It does nothing to make an install cheaper.

What the breakout launches did differently#

Moloco's study of 55 new mobile games looked at what separated the breakouts from the rest. One difference was creative volume.

Distinct ad creatives per launch, first yearBreakout launches against other launches in Moloco's sample of 55 new mobile games

93Breakout launches

37Other launches

Distinct ad creatives per launch, first year

Breakout launches	93
Other launches	37

The timing matters as much as the count. The breakouts introduced 47% of their creatives in months 7 to 11, against 27% for the others. They kept feeding the ad auctions new work for most of the year, not only at launch.

Run the numbers on a breakout. 47% of 93 is about 44 creatives across five months: roughly 9 a month, or 2 a week, in the second half of year one. If each goes out in three sizes (feed, story and landscape), that's around 130 files in five months, while the launch work still needs refreshing. That's a production plan, and in our experience few small teams budget for it before launch.

The biggest advertisers work at a different scale again: AppsFlyer puts the top gaming advertisers at 2,400–2,600 variations every quarter. AppsFlyer's Adam Smart summed up the market as "not a shortage of creativity, but a surplus of it." When everyone ships more ads, a game with a launch batch and nothing after it falls behind in the auction.

The caveat is real: bigger launches have bigger budgets, so creative volume is partly a symptom of success. But no studio gets to 93 creatives by accident. It needs a way to make them.

If you make games, plan the players before the build#

Prompt-to-game tools will make prototypes cheaper, and that's good news for anyone testing ideas. It doesn't change what decides whether a game finds an audience. Here is what we'd do about it.

1. 1
   Plan for players while you're still building

   Prompt-made prototypes make building the cheap step. Decide early who the game is for, which networks you'll buy on and which hook you'll lead with.

2. 2
   Budget creative for a year, not a launch

   Moloco's breakouts introduced almost half their creatives in months 7 to 11. Plan a steady flow of new concepts and variations, not one launch batch.

3. 3
   Make your game's art reusable for ads

   Characters, backgrounds, logo and UI as clean, separate files. Every ad you make later is assembled from them, so this is the cheapest step to do early.

4. 4
   Test hooks before you scale spend

   Most of an ad's job happens in the first seconds. Test several openings per concept, then make variations of the one that wins.

For the wider playbook, see how to market a mobile game. For the openings, see how to test the first 3 seconds of a game ad.

The production side is where we work. A hundred-odd files in five months is a lot to build by hand for a small team, and every one has to look like the game. Vidscape turns the art a game already has into layered mobile game ads, so a new character, headline or size is an edit rather than a rebuild. If you're a small team, start with Vidscape for indie studios.

Frequently asked questions#

Sources

1. Introducing Playground: Create and play custom games — Google · Oct 2026
2. Google experiments with an AI-powered gaming platform — TechCrunch · Oct 2026
3. Google Labs Playground — 9to5Google · Oct 2026
4. Google Playground and Unity Spark will enable creators to easily generate 3D games — GamesBeat · Oct 2026
5. Google officially reveals AI game-creation platform Playground — VGC · Oct 2026
6. Google unveils Playground to create games from text prompts — Inven Global · Oct 2026
7. Build without limits on Roblox — Roblox · Jul 2026
8. RDC 2026: The world needs more play — Roblox · Sep 2026
9. New Steam releases hit record high of 720 in one week — Inven Global (ICO Partners data) · Sep 2026
10. Steam data shows over 19,000 games released in 2025 — KitGuru (SteamDB data) · Dec 2025
11. H1 2026 Digital Gaming Market Index — Sensor Tower · Jul 2026
12. Steam Replay: only 14% of playtime was for 2025 releases — GamingOnLinux (Valve data) · Dec 2025
13. App and game store statistics — 42matters · Oct 2026
14. Only 2,500 of 190,000 mobile games released in 2025 surpassed 500,000 downloads — PocketGamer.biz (Moloco report) · Aug 2026
15. State of Gaming App Marketing 2026 — AppsFlyer · 2026
16. AppsFlyer gaming marketing report press release — AppsFlyer · Jan 2026
17. Adjust Gaming App Insights report (summary) — GameDevReports · Apr 2026
18. AI, creative velocity and player intelligence are reshaping mobile game marketing (SocialPeta H1 2026) — Game World Observer · Sep 2026
19. Games and AI with Joost van Dreunen — Mobile Dev Memo · Mar 2026