Why Grow a Garden and Steal a Brainrot Bet on Boring Idle Tech: The Luau Loop Behind 25-Million-Player Records
Two of the ugliest games of 2025 beat everything beautiful. A farming plot where carrots grow while you are offline, and a concrete room where cartoon sharks steal your money, drew more concurrent players than Fortnite and Steam combined on the same weekend.
The short answer is technical, not aesthetic: both games run on Roblox Studio with tiny replication payloads, offline timers, and appointment-based scarcity. They treat other players as content and waiting as gameplay. That boring stack scales to tens of millions where AAA density would collapse.
The records are real, and they are idle games
Grow a Garden released on Roblox on March 26, 2025 as a free-to-play multiplayer idle game. Players buy carrot seeds with Sheckles, plant, harvest, and buy progressively exotic seeds. Crops continue to grow while the player is offline, seeds go in and out of stock, and weekly exclusive items are only available if you are online to claim them.
The numbers are absurd for a game its original developer built in three days:
- over 5 million concurrent users (CCU) on May 18 with the Blood Moon update, nearly doubling the prior Roblox record held by Blox Fruits
- over 11.7 million CCU on May 31 with Bizzy Bees, over 16 million CCU on June 14 with Working Bees — surpassing Fortnite's 15.3 million CCU record from December 2020 and the peaks of PUBG and Counter-Strike 2 on Steam
- 21.3 million CCU on June 21 Summer Update, then at least 22.3 million CCU on August 23 during a staged admin war
- 35.3 billion visits and 10.9 million favorites as of May 2026, fastest Roblox game to 1 billion visits in 33 days
Steal a Brainrot released May 16, 2025, also as a Roblox Studio idle multiplayer game. Players buy Brainrots — voxel characters based on Italian brainrot memes like Tralalero Tralala — off a central conveyor belt. Owned Brainrots generate income passively. You can also steal them from other players' bases, in a loop compared to capture the flag.
Its curve is even steeper:
- 5 million peak CCU in July 2025, 20 million peak CCU in August 2025
- second Roblox game to pass 20 million CCU after Grow a Garden, then 24 million CCU on September 13 Extinct Event, then 25.4 million CCU in October — the first game ever to pass 25 million concurrent players
- on August 23 the joint admin war pushed Roblox itself to 47.4 million platform CCU, beating its July record of 32.69 million
Independent reporting confirms the scale. IGN reported that on August 23 Roblox hit 47 million players while Steam drew 37.9 million online accounts with 11.2 million in-game and Fortnite drew a 1.6 million daily peak. On that day Grow a Garden drew 22 million and Steal a Brainrot drew 15 million individually.
Both games are now partly owned by DoBig Studios, which also holds a minority share in Grow a Garden alongside Splitting Point Studios and original creator BMWLux. That ownership matters because both games use the same FOMO playbook: limited skins, battle passes, and weekly admin-abuse events.
What the loops actually do
Grow a Garden is FarmVille stripped to a timer. You start on an empty plot among other players' plots. Plant, wait, harvest, reinvest. Pets come from loot boxes. Robux speeds up waiting, adds perks, and lets you steal crops. The visuals use studded textures evocative of old-school Roblox, and the soundtrack is classical music including Mozart's Rondo Alla Turca.
Steal a Brainrot is harsher. As Polygon's explainer documents, the map is only cage-like houses and a line of Brainrots marching down the middle. You start with a little cash, buy off the belt or steal immediately — stealing a rare Brainrot is more profitable than grinding from zero. Bases display collections on pillars, you cannot decorate, and defense is a single shield button that lasts 60 seconds before cooldown. Owned Brainrots pay forever until stolen. Rebirth wipes everything for better stats and more money.
That meanness is the point. Screen Rant calls it addictive because of lack of complexity and a straightforward loop. Polygon notes the best items are cash-only, from sub-$5 guns to $49.99 server-admin abilities — pay-to-win wrapped in a game literally named after theft.
The boring tech that lets this scale
Both games list Roblox Studio as engine. The scripting language is Luau, described in Roblox Creator Hub docs as a fast, small, safe, gradually typed embeddable language derived from Lua 5.1.
That choice is load-bearing:
Small state, cheap replication. A carrot timer or Brainrot income tick is a number and an owner ID. No skeletal mesh, no physics authority, no NavMesh. When you need to shard 20 million players across thousands of 30-player servers, tiny state wins.
Offline progress as database, not simulation. Grow a Garden crops grow while offline. That is not a running server burning CPU — it is a timestamp and a growth curve evaluated on return. Idle games turn wall-clock time into free content.
In-stock / out-of-stock and weekly exclusives as scheduler. Seeds going in and out of stock and admin-abuse exclusives create synchronized login spikes. PocketGamer observed Grow a Garden does not sustain peaks — CCU fell to 7.6 million within an hour after a 21.3 million spike. That volatility is intentional: spikes drive store conversion, troughs save server cost.
Stealing as server-paid content. Capture-the-flag theft means players generate drama for each other. No AI director, no quest scripting. The 60-second shield plus visible wealth (you can see into bases to price targets) creates constant low-grade surveillance — the Gen Z stare as game mechanic.
The browser parallel runs on Unity WebGL. A Unity Web build ships a Build folder with a JavaScript loader, framework runtime, WebAssembly binary, and asset data, compressed with Gzip or Brotli for instant load. EasyHub's own pitch is hundreds of HTML5 and Unity WebGL games that load instantly with no downloads. That is the same idle promise in a tab: no install, timer-driven progress, prestige reset.
If you want to feel the loop without opening Roblox, two browser examples make the mechanics tactile. Backpack Idle is an idle game where you drag artifacts into backpack slots to maximize earnings, research unlocks, and prestige to restart stronger while keeping passive income even when AFK — the same faucet-and-prestige math as Grow a Garden's seeds and Steal a Brainrot's Rebirth, but with local inventory instead of server theft. Brainrot Merge: Drop Puzzle and its parent Italian Brainrot Merge distill the other half: drop and combine meme animals like Tralalero Tralala and Cappuccino Assassino before the board fills, which teaches merge-tension in 90 seconds that Roblox stretches over weeks. For pure theft anxiety, Brainrot Idle Clicker taps through bizarre enemies and card evolutions — clicker income with stranger powers, no base to defend.
My take: ugliness is infrastructure, friction is retention
Most coverage stops at simplicity or calls these games prototypes. Kotaku said Grow a Garden looks more like a prototype than a finished game. PCGamesN called it a pretty standard AFK game that is hard to put down. That misses why prototype aesthetics are an advantage.
Here is my contrarian read, labeled as interpretation rather than verified fact:
1. Low fidelity lowers the fairness bar. If a photorealistic game let paying players steal your $50 shark, you would call it fraud. When a voxel shark with Nike shoes steals your voxel shark, you laugh, clip it, and post the kid crying on TikTok — 36.2 million views on one Steal a Brainrot crying video. Ugliness makes exploitation feel like a joke, which makes pay-to-win and theft socially transmissible. Polish would make the same economy feel insulting.
2. Waiting is cheaper than playing. AAA games sell moment-to-moment fun and pay server cost every second you are engaged. Idle games sell return-to-moment fun and pay almost nothing while you are gone. Offline growth plus stock rotation turns absence into anticipation. The game does not need to be fun for 3 hours; it needs to be worth checking for 3 minutes, 10 times a day.
3. Scarcity beats balance. Neither game is balanced in a competitive sense. Limited-time items, out-of-stock seeds, and conveyor rarity mean you miss out unless you are there. DoBig's reputation for FOMO monetization is criticized by players, but from an engineering view it is load management: concentrate demand into predictable windows, monetize the spike, let troughs cool the fleet.
4. Kids are the anti-botnet. TD Cowen alleged bot inflation from Philippines and Indonesia account creation, then retracted manipulation claims while noting the upsurge. Roblox countered that preliminary analysis confirms genuine popularity. My read: 35% of Grow a Garden players under 13 and ~40% of Roblox users under 12 are not a weakness — they are the hardest audience to fake at scale because they bring parents' devices, school schedules, and TikTok virality. Bots do not cry on camera. Social video is the proof-of-humanity layer.
Unity WebGL idle games prove the same loop works without Roblox. They swap server-side DataStores for localStorage, and admin wars for daily rewards, but keep the core: faucet, timer, prestige. The difference is trust. Roblox can enforce theft and trading; a browser tab cannot, so it leans on merge and inventory optimization instead. That is why EasyHub's idle and merge catalog feels like a lab version of the Roblox hits.
What indie devs can actually steal
Do not clone Brainrots. Clone the cost structure:
- Build the faucet first, the world second. One number that goes up offline, one sink that resets it (seeds, Rebirth, prestige). If that 2-button loop is not compelling in a spreadsheet, art will not save it.
- Add one social tax. Theft, visiting fees, leaderboard theft, or shared stock. One player must be able to meaningfully set back another in under 60 seconds, with a visible counter (shield timer, stock countdown). That is your content pipeline.
- Schedule scarcity, not content. Weekly exclusives and rotating stock beat new maps for retention per engineering hour. Two small teams on rotating events, as Grow a Garden did in July, beats one big expansion.
- Ship ugly on purpose. Studs, voxels, and single-screen bases replicate cheaply and read instantly on phones — where most Roblox growth comes from. Spend the saved budget on live-ops tooling: admin commands, event flags, and rollback for dupes and script cheaters, which GameRant notes already plague Steal a Brainrot via cash-collection automation.
Start on the web to prototype the math. A WebGL idle with Brotli compression loads in seconds and tests faucet pacing before you pay Roblox server and moderation costs. Then move to Roblox Studio and Luau if you need theft, trading, and viral social distribution.
Bottom line
Grow a Garden and Steal a Brainrot did not win despite boring tech. They won because boring tech — Luau timers, tiny replicated state, offline timestamps, rotating stock, and player-as-content theft — scales to 22–25 million concurrent players while beautiful tech struggles to fill a 100-player lobby.
The lesson is uncomfortable: in 2025, the most powerful engine feature was not Nanite or ray tracing. It was a clock that kept paying you when you logged off, and a neighbor who could take it all when you logged back on.
Top comments (0)