Direct answer — Why are mobile game downloads flat while short-drama apps and AI assistants are growing faster than games in 2026? Mobile game downloads are not flat, they are shrinking. Sensor Tower’s State of Gaming 2026 puts 2025 mobile game installs at 50 billion, a decline of 7% year on year, while total App Store and Google Play downloads across every category edged up 0.8% to nearly 150 billion. That gap is the most consequential of the mobile game industry trends 2026 has produced: short-drama apps and generative AI assistants are absorbing install and attention growth that games used to capture by default, and consumer spending on non-game apps overtook spending on games for the first time. Games did not stop making money in 2025. They stopped growing automatically.

Calling mobile game downloads flat in 2026 is the polite version. The market did not contract, but games lost their default share of it, and that is a strategy problem, not a marketing problem. Across 20+ years at Gameloft, SFR, Blacknut and Impulse Media Hub, roughly €12M+ of P&L and 50+ launches, I have watched studios answer a flat install curve by spending more on user acquisition. That works in an expanding market and fails in this one. What follows is a decision guide: what the 2026 data means, how to tell whether your studio sits in the flat cohort, and what to stop funding before you fund anything new.

The headline is a divergence. Total mobile app activity grew, mobile game activity did not, and the money moved with the attention.

Metric (2025, App Store + Google Play)FigureYoY changeReported by
All-category app downloadsNearly 150 billion+0.8%Sensor Tower, State of Mobile 2026
All-category IAP and paid app revenue$167 billion+10.6%Sensor Tower, State of Mobile 2026
Non-game IAP revenueSurpassed games for the first time+21%Sensor Tower, State of Mobile 2026
Mobile game installs50 billion-7%Sensor Tower, State of Gaming 2026, via PocketGamer.biz
Mobile game IAP revenue$82bn+1.4%Sensor Tower, State of Gaming 2026, via PocketGamer.biz
Generative AI app downloads3.8 billionDoubledSensor Tower, State of Mobile 2026
Generative AI IAP revenueAbove $5 billionNearly tripledSensor Tower, State of Mobile 2026

Three things follow, and none are cosmetic.

First, the games market is still growing, just barely. Revenue growth of +1.4% year on year is not a crisis, but it is a rate at which no venture-backed growth plan survives contact with a board. Sensor Tower’s State of Gaming data also puts average in-app purchase revenue per download at $1.62 in 2025, the number that matters most when install supply shrinks: growth now comes from value per player, not player count.

Second, distribution concentration did not loosen. Google Play accounted for 81% of all game downloads in 2025, with the App Store contributing 15%, and free-to-play made up 96% of downloads overall and 99% on Google Play. If your plan assumes a meaningful escape from those two storefronts, it is a five-year plan, not a 2026 plan.

Third, retire the phrase growth at all costs. Deconstructor of Fun’s read of the same dataset is blunt: the era of growth-at-all-costs is over. They also note that games had a quieter year with no new $1 billion entrants, so the top of the market stopped minting winners, not just the middle.

Short Drama Apps vs Mobile Games: Where the Attention Went

Short-drama apps are the clearest example of a category that took what games assumed was theirs. Sensor Tower reports that global short drama downloads surpassed 850M in Q1 2026, up 140% year on year, with in-app purchase revenue reaching roughly $750M in the quarter, up 20% year on year. Spend growth is moderating while install growth is not: the profile of a category still buying its audience.

The threat is not that short drama is a better product, but that it monetises narrative attention at a fraction of a game’s production cost and time-to-market. Sensor Tower’s analysis of the category found meaningful overlap with gaming audiences, with games making up over 30 of the top 100 apps most used by short-drama audiences in the US and more than 50% in Brazil during 3Q24. In LATAM, the same analysis notes that short-drama apps have already outperformed several newly launched mobile games by revenue.

Generative AI assistants apply the same pressure from a different direction. Time spent in generative AI apps reached 48 billion hours in 2025, roughly 3.6x the 2024 total, and ChatGPT alone generated $3.4 billion in annual in-app purchases according to Deconstructor of Fun’s analysis. Meanwhile, average time per user sits at 3.6 hours per day and has barely budged. AI assistant apps growth in 2026 is therefore not additive to the attention pool. It is redistributive.

That is the reframe I push in every gaming growth advisory engagement this year: you are no longer competing for a share of a growing attention market, but for a share of a fixed one, against categories with shorter production cycles and faster payback.

Where does your studio actually stand? If your install curve has flattened and you are not sure whether it is you or the market, book a strategy call and we will separate the two before you spend another euro on user acquisition.

Are You in the Flat Cohort? A Four-Question Diagnostic

Market-level data is useless until you can locate yourself inside it. Run these four checks before changing anything.

  1. Is your install decline steeper than -7%? The market moved -7% in 2025. If you are at -5%, you are outperforming a shrinking category. If you are at -25%, this is a title-level problem wearing a market-level costume.
  2. Is your revenue per install rising? If installs fall and revenue per install rises, your monetisation is doing its job in a maturing market. If both fall, you have a product-market fit problem that user acquisition budget will only disguise.
  3. Is your paid user acquisition still the majority of your installs? Paid channels are where the squeeze is sharpest, so understand why mobile CPIs keep climbing even as AI floods creative pipelines before concluding your media buying team is underperforming.
  4. Do you know your per-cohort platform take rate? After the 2026 storefront changes, the effective fee on a new install and an existing install are not the same number. If you are still modelling a single blended rate, your unit economics are wrong before you start.

If two or more of those answers are bad, you are in the flat cohort. That is not fatal, but the growth plan you wrote in 2023 is now actively misleading.

The 2026 Reset: What to Stop Funding and What to Fund Instead

A reset is a reallocation, not a rescue. Here is the shape it takes.

Growth-at-all-costs default2026 resetWhat changes in practice
Buy installs to hit a volume targetBuy installs to hit a payback targetVolume becomes an output of unit economics, not an input to the plan
One blended platform fee in the modelPer-cohort fee modellingNew installs, existing installs, and direct channels are priced separately
Roadmap built around new-user featuresRoadmap built around depth for the existing baseLive operations and retention work outrank onboarding polish
Treat other games as the competitive setTreat attention categories as the competitive setShort drama, social, and AI assistants enter the session-time analysis
Growth is a marketing functionGrowth is a P&L functionThe decision owner moves from performance marketing to leadership

Two lines deserve specific attention. Per-cohort fee modelling is not theoretical: the 2026 changes to app store platform economics split the platform service fee from the billing fee and made the effective take rate depend on install cohort and payment route. Studios still modelling a flat rate are budgeting against a number that no longer exists.

The second is the competitive set. Most studios I work with still benchmark session length against other games in their genre. That was the right frame when games competed with games. When your churned player’s next session is a short-drama binge or an AI assistant, genre benchmarks tell you nothing about why they left.

Once the diagnosis is done, the execution layer is well-trodden ground, documented separately in the complete mobile game growth playbook. This article is deliberately not that playbook: running better tactics against the wrong strategic assumption is the most expensive mistake available in 2026.

What We Could Not Verify

Full transparency on the limits of this analysis.

  • Full-year 2025 short-drama revenue. Figures of 115% year-on-year growth and USD 2.98 billion of 2025 in-app revenue circulate widely in secondary coverage of Sensor Tower’s work. We could not confirm them on a primary Sensor Tower page, so this article uses only the Q1 2026 figures Sensor Tower publishes directly.
  • Two Sensor Tower reports, two growth rates. State of Mobile 2026 and State of Gaming 2026 publish slightly different 2025 growth rates for mobile game in-app purchase revenue. We use the State of Gaming figure throughout and did not average or blend the two.
  • Genre-level install losses. We could not find a verified breakdown of which mobile game genres lost the most installs in 2025, so this article makes no genre-specific install claims.
  • Substitution versus incrementality. The audience overlap Sensor Tower reports between short-drama and gaming audiences is from 3Q24 and describes overlap, not substitution. We do not claim a causal transfer of time from games to short drama.

The Strategic Read

The mobile game industry trends 2026 has surfaced are not a downturn. They are a change of regime. Downloads stopped being free, attention stopped expanding, and the growth premium moved to categories that convert attention into revenue faster than games do. A studio’s job now is to know exactly where its numbers sit against the market, then reallocate deliberately instead of spending harder against a curve that no longer bends.

Ready to pressure-test your 2026 growth plan against the market data? Start with how Game Growth Advisor works with studios, then book a strategy call for a straight read on whether your flat installs are a market condition or a fixable one.