Unpredictable Hits #012: 262 Games Launched, 3 Won + What AI Changed in UA Creative
From the Editor
What's up friend?
Gaming just closed its first $200 billion year. Downloads fell.
Those two facts are the whole industry in two sentences. Mobile did $113 billion of that record, up double digits, while downloads shrank. Growth now comes from squeezing more out of the players we already have.
Meanwhile every new install costs 30% more than it did a year ago, and the paid-to-organic ratio jumped 61%. Nobody is coming to save your UA budget with cheap users.
This issue is about where winners actually get made. We published a piece on what AI really changed in UA creative (production, not winning). 262 casual games launched in Q1 and three of them matter. A merge game out-earned Candy Crush. A Turkish studio you know hit a $500M run rate in two years by killing most of its own ideas. And the best-performing UA channel in mobile gaming is the one almost nobody funds.
Let's go.
Featured: The State of AI in UA Creative
We wrote down our honest read on what AI actually did to UA creative. The short version: AI nearly solved making more. The winners part is still on you.
What changed: Production. More than half of top-grossing mobile games already run AI somewhere in their creative pipeline, and the cost of another variation keeps falling.
What didn't: The hit rate. Game launches roughly doubled and the number of hits barely moved. We got much better at making content and no better at finding the ones that win.
Why creative carries all the pressure: ATT rotted the targeting data, and the networks pulled the remaining dials (bidding, budgets, audiences) inside their own models. Creative is the one lever still in your hands. AI arrived exactly when the demand for volume became bottomless, which is why production got automated first.
The real job: Finding a winner is a loop. Decide what to try, build it, read what happened, repeat. AI is cheapening the middle step, which was always the easiest. The other two are judgment. You can make a thousand ads a day and be no closer to knowing which one to make next.
The question that sorts every tool: Does it close the loop, and does it close it on my game? A tool that only generates saves you money and wins you nothing that lasts. In our experience, about 97% of attempts fail. The loop is how you find the 3%.
Industry Buzz
1. Gaming Crossed $200B. The Growth Isn't Coming From New Players.
Author: Michiel Buijsman, Newzoo
The global games market closed 2025 at $201.6 billion, up 9.1%, crossing $200B for the first time. Mobile did $113.3 billion of it, up 10.7%, and now holds 56% of the market.
The twist: mobile grew while download volumes declined. Newzoo calls it โa market growing through deepening monetization rather than expanding reach.โ Player counts are flat. Spend per player is doing all the work.
The UA numbers hurt: Cost per install rose 30% in 2025 to $0.56 globally, and the paid-to-organic install ratio jumped 61% (per Adjust). Organic discovery keeps eroding, and you pay more for every user who replaces it.
The forecast: $234.4 billion by 2028, with mobile at $121.1 billion in 2026. GTA VI is the year's defining catalyst on the premium side. On mobile, D2C keeps compounding: Appcharge hit $1B in annualized transaction volume in March, double where it was last July.
When reach stops expanding, extraction wins. And extraction starts with creative that converts the users you can still afford.
2. 262 Casual Games Launched in Q1. Three Matter.
Author: Matthew Emery, Gamigion (data: AppMagic)
Gamigion let Q1's casual launches bake for three months and counted the survivors. 262 games shipped. Three of them, 1.1%, are running at $10M+ a year. Thirteen cleared $1M. The other 249 haven't.
The winner: Hollywood Merge by Yotta, alone at a $64M run rate. Nothing else came close: Hotel Legacy: Merge at $16M, then Hatch Dragons at $11M from tiny NZ studio Runaway Play, the quarter's one genuine surprise.
The genre story: Merge-2 owns new casual releases in 2026. Three of the top four and four of the top ten are merge games, and merge took $66M of the cohort's ~$121M total revenue. That's 55% of the money from a fraction of the launches.
The uncomfortable read: Puzzle made up over half of all launches, and non-merge puzzle had a 0.0% hit rate into the $10M tier. The industry's most crowded lane produced zero big winners outside merge.
The odds don't care about your conviction. You can stack them with genre and craft. After that, it's exploration at volume.
3. A Merge Game Out-Earned Candy Crush. The Reason Is Structural.
Author: Aylin Yazฤฑcฤฑ, Deconstructor of Fun
In February, Gossip Harbor did $77.7M against Candy Crush Saga's $71.7M. A merge game from Chinese studio Microfun out-grossed the most famous puzzle game on earth in a single month. Gossip Harbor grew 172% year-over-year. Candy Crush fell 7%.
Yazฤฑcฤฑ's explanation is the best genre analysis I've read this year. Match-3 boards are temporary: every level resets, and a failed level gives you a clean exit. Merge boards are persistent: your orders, generators, and half-finished chains sit exactly where you left them, forever.
The line that captures it: โMatch-3 sells you a way out of a problem, while Merge sells you a way to keep going on something you already started.โ And the kicker: โFailing gives the player an exit. Pausing doesn't.โ
The verdict on the title question: Merge earns more per player. Match-3 still earns more in total because its audience is far bigger. But the trend has picked a side: the Merge-2 genre grew 80% in 2025 while long-running match-3 subgenres lost both revenue and downloads.
The stat buried in the charts: Merge-2 pulled 124 new complex-meta attempts in 2024. Two succeeded. A 1.6% hit rate in the hottest genre in casual. Even when you pick the right wave, you still have to survive the water.
๐ Read on Deconstructor of Fun
4. Grand Games: $0 to $500M Run Rate in Two Years
Podcast: Michail Katkoff, Deconstructor of Fun, with Bekir Batuhan รelebi, Co-founder & CEO of Grand Games
Two years ago Batuhan รelebi had 20 people in a 70-square-meter office in Istanbul, hand-designing levels for Magic Sort while raising money. Today Grand Games runs a $500M annual run rate on $103M raised, with revenue up fivefold in the last year. Katkoff's read: โThat trajectory doesn't exist in mobile gaming right now.โ
The structure: Five autonomous studios of roughly 15 people each. Every head of studio owns design, development, and live ops end to end. Nobody asks the founders before shipping a feature. Above 20 people, Batuhan found, the job shifts from product management to people management and ownership evaporates.
The greenlight filter: If there's no impactful differentiation in art, level design, or core product, the game doesn't begin. โMaking an existing game slightly better isn't a strategy.โ Car Match got greenlit at month three. Then Grand went almost a full year without greenlighting anything. Katkoff calls it the sniper strategy in a market where everyone blasts away with shotguns.
The self-description: โWe're kinda fast with kinda good quality.โ Deliberately parked between Dream Games' perfectionism and Rollic's volume machine. Fast enough to test real hypotheses, good enough that quality doesn't poison the signal.
The honest gap: Batuhan admits their hit titles are under-optimized. Zero-to-one and one-to-a-hundred are different muscles, and Grand built the first one. Worth remembering when you staff your own teams: the people who find winners are rarely the people who squeeze them.
๐ Listen on Deconstructor of Fun
5. Rewarded UA: $77 Per Install While Everyone Looks Away
Author: Matej Lancaric, covering the first Almedia Summit in Berlin
Matej spent two days at Almedia's first summit (Almedia runs FreeCash, ~$1B in UA spend, roughly 30% of the rewarded market) and came back with AppsFlyer's on-stage numbers. Rewarded UA generates $77 in revenue per install. Ad networks average about $8. SRNs about $12. ROAS: 8.11x on rewarded against 4.49x on SRNs.
Matej's reaction: โRead that again. Not 10% better. Not 'promising early signals.' Six times the revenue per install of your average ad network.โ
The gap: Despite those numbers, rewarded moved from just 5.9% to 6.7% of gaming UA spend in a year. Eric Keen of Supersolid admitted he spent two years chasing Unity, Google, and Apple before realizing his games were far stronger on rewarded. Aviad Biton of Whalo put his cost of waiting at $3-4M.
The catch: Rewarded is its own discipline, and treating it like another line in your MMP dashboard is how teams burn money on it. The panel's playbook: model your in-game economy first, start on CPE to learn the funnel, move to CPI to scale, and judge the ROAS-to-CPI relationship instead of chasing cheap installs. A $25 CPI can out-earn a $12 one on the same network because bigger rewards raise intent.
The creative angle: Pixel Federation's Michal Bubernik went from 3 cinematic videos a month to 300+, peaking at 700, with the same 10 artists. Iterations that took 3 days to 3 weeks now take under 30 seconds. Matej's summary: โIf you don't have a large creative volume, you are already losing.โ
6. AI Money Is Flooding Into Games. Follow Where It Lands.
Author: Naavik's news roundup
Three AI stories from Naavik's latest issue that mobile teams should actually care about.
Kinoa raised $10M to point AI at the operations layer: predictive agents that anticipate churn, spot future spenders, and surface the right offer without a code release. Built by Playtika, Amazon, and Skai veterans; customers include Playstudios and Playsimple; claims over 25% revenue lift across its customer base. Naavik's framing nails the timing: โmobile user acquisition costs have increased while targeting efficiency has declined, squeezing the economics that mobile publishers have relied on for years.โ
General Intuition raised $320M at a $2.3B valuation to train frontier AI models on gameplay data from Medal's 17M+ monthly active players. Gaming data is now valued as foundational training material for the broader AI economy. If your players generate spatial reasoning data, someone may eventually pay for it.
EA says AI is delivering โfaster prototypingโ and a real rise in creativity across its studios. Naavik's response is the right one: โNo shipped titles were referenced, no metrics were shared, and the specific types of AI being used were left undefined.โ Executive AI enthusiasm is free. Shipped results are the test.
๐ Read the full Naavik issue
7.World Models: Gaming Is Sitting on the Data Everyone Wants
Author: Matthew Emery (LinkedIn)
More than $10 billion flowed into world model companies over the past 18 months, and Courset wrote the clearest breakdown I've seen of what it actually means for game companies.
The core insight: world models learn cause and effect, and that can't be scraped from the web. Gaming sits on the richest supply of it through gameplay video and engine-level capture. In his words, this โreframes data from a supporting asset into a core value driver for gaming companies.โ Medal's roughly two billion clips a year is why General Intuition raised $134M.
What it means for studios: a talent arms race first, since the people best equipped to build world models โmay already be sitting in game studios,โ and acqui-hires keep pulling them away. Then the first real payoff, pipeline efficiency: drafting environments, scenes, and storyboards on demand, so teams iterate on atmosphere and pacing before committing to full asset production. The consumer-facing dream-worlds stay far out. Genie 3 inference still runs roughly $100 an hour, and games need stable, repeatable rules that dreamed worlds don't deliver yet.
The grounded close: the billion-dollar rounds aren't underwritten by gaming. Gaming's data may end up catalyzing everything else, the same way 3D graphics and online play did.
Wrap Up
If you zoom out for a second, everything in this issue is describing the same shift from a different angle. The market keeps growing, but it grows by getting more out of the people who already play, and the price of reaching anyone new keeps climbing. At the same time, making things has never been cheaper. Games, ads, creatives, all of it pours out faster than the market can absorb, and most of it disappears without leaving a mark.
That combination quietly changes what a good team looks like. Grand Games wins by refusing to build anything that doesn't have a real reason to exist. Pixel Federation wins by letting the same ten artists move at machine speed. The rewarded UA crowd wins by putting budget where the returns actually are, while everyone else keeps feeding the same three networks out of habit. None of them out-produced anyone. They got better at finding what works, and they had the discipline to follow it wherever it pointed.
That's the whole game now. You can't predict a hit. You never could. But you can be the team that finds yours first.
See you next time.
Yours, Jonathan (just got back from an island in Greece, no AI agents were to be found) Fishman
About the Author
Fishi is the Head of Marketing at Sett. His brain is a chaotic jukebox of ideas with more cultural references than any feed can handle. He collects sneakers and plays chess while youโre still counting sheep.
