Roblox Can't Prompt Its Way Out of the Viral Problem

Roblox is holding its annual developer conference this week in San Jose, showing off AI tools that turn text prompts into playable games, short-form video feeds, and real-time voice translation. The stock has fallen 70% from its peak.Shares dropped 74% from peak.

The product announcements are interesting. They are not the point.

The point is the number the product announcements are trying to fix: Roblox daily active users dropped from 152 million to 123 million in six months. And bookings — the actual cash that flows through the platform before it is recognized as revenue — grew just 8% in the second quarter of 2026, and are expected to decline 14% to 18% next quarter. This is the first time the company has guided to negative bookings growth.

To understand why, you need to look at what actually drove those 152 million users.

The viral game problem

In the summer of 2025, Roblox experienced a kind of platform-wide frenzy. Games like "Grow a Garden" and "Steal the Brainrot" became cultural moments. They pulled in millions of daily players, and those players spent money. The quarterly peak — 152 million daily active users, 40 billion hours engaged — was real. It was also accidental.

Nobody designed it. Nobody could replicate it. A small number of creators made games that kids could not stop playing, and the rest of the platform rode along.

Then the kids moved on. That is what kids do. "Grow a Garden 2" exists. Nobody is talking about it.

The decline from peak was not entirely natural. Roblox also changed its discovery algorithm to recommend less viral content, reasoning that short-term viral spending would give way to sustainable engagement. Management also shut down the sale of cross-experience game passes, a move that hurt near-term monetization. These are rational decisions in the abstract. Together they mean the company is currently navigating a period where both the organic engine and the monetization engine have been downshifted at the same time.

Revenue does not tell the whole story

Roblox reported $1.5 billion in revenue for Q2 2026, up 36% year-over-year. That sounds strong. Revenue on a platform like this is recognized as users spend their virtual currency over time, which creates a lag between what people actually spend and what the income statement shows.

Bookings remove that lag. In Q2, bookings were $1.6 billion — up only 8%, at the low end of guidance, far below the 70% bookings growth the company posted just two quarters earlier. For Q3, bookings are expected to fall to between $1.58 billion and $1.65 billion, below Wall Street's already lowered estimates. The company also withdrew its full-year guidance, citing "higher variability."

Meanwhile, the average amount each daily user books has declined. In the U.S. and Canada, it fell 5% year-over-year to $38.63. The under-13 demographic, which matters most for a platform whose users eventually age out, saw the biggest monetization drop.

You can look at the international numbers and find comfort there — Japan DAUs up 67%, India up 64%. But those users spend far less per day. The APAC bookings per user at $4.90, compared to nearly $39 in North America. User growth at lower monetization levels is a real trend, but it does not replace the loss of high-spending engagement at home.

The product answer

So Roblox is building more tools. The main one is "Build," an AI-powered creation feature inside the mobile app. Type a sentence — "a cozy adventure game set in a dense forest with environmental obstacles" — and it generates a starting project with mechanics, characters, visuals, and sound. The public alpha launched in New Zealand in late July 2026. The idea is that millions of players who never opened Roblox Studio on a desktop can now become creators.

The company also launched "Moments," a TikTok-style short-form video feed for gameplay clips, hoping to turn highlight reels into discovery and engagement. It has been expanding AI tools for voice translation, 3D object generation, and automated playtesting.

The stated ambition behind all of this is audacious: Roblox wants 10% of all gaming content revenue to flow through its platform. The global games market reached roughly $200 billion in 2025. That would mean Roblox processing $20 billion in bookings a year, compared to about $7 billion for all of 2025.

That is not unreasonable as a long-term aspiration. The question is whether AI-generated games are the bridge.

The gap between creating and compelling

Roblox has always sold itself on the idea that "you make the game." The truth was always more constrained: a very small number of people make the games that drive the platform's engagement and spending. Most users never create. The ones who do create rarely build something millions play.

AI lowers the barrier to creation dramatically. But making a game is not the same as making a game people want to play and spend money in. The games that drove Roblox to its peak were not the most technically sophisticated. They were culturally resonant. They fit a mood that existed among millions of kids at a specific moment.

There is no prompt for that.

This is not a criticism of the technology. Build and Moments are exactly the kind of bets a platform should make — open the creator funnel, give users more reasons to stay. The risk is more structural: if Roblox's growth depends on unpredictable viral phenomena, and the company's response is to manufacture more content more efficiently, then it may be solving a supply problem when the real issue is demand.

More games does not guarantee more engagement. If anything, the opposite may happen — more content could fragment attention further, making it harder for any single game to achieve the kind of dominance that pulls in millions of daily players.

The test

The stock trades at roughly $32 billion in market capitalization. The company has about $6 billion in cash. It is not in immediate danger. The question for investors is whether the current trajectory represents a cyclical trough or a structural change in how kids engage with the platform.

Here is a simple way to think about it. Roblox has a boom-bust pattern driven by viral games. If the troughs are rising — if each new low is higher than the last — then the company is growing even during its quiet periods, and the current decline from peak may be a temporary correction. If the troughs are falling, or flat, then the viral hits were not anomalies on a rising trend, but the entire engine.

The second-quarter number — 123 million DAUs, down from the 132 million reported in Q1 — puts the next earnings report in January 2027 at the center of the argument. If DAUs continue to fall through the holiday season, when engagement should naturally spike, the market's 70% drawdown may have been too small, not too large. If users stabilize and bookings stop declining, the company has room to recover.

The product announcements at the developer conference are the company trying to influence that outcome. Whether they succeed will not be clear from what gets announced. It will be clear from what kids choose to do on a Thursday afternoon.