If you've shipped an app in the last year, you've felt it: the build was faster than ever, and the launch was quieter than ever. That isn't bad luck. It's the new shape of the market, and the numbers behind it are worth understanding before you plan your next quarter of creative.
The supply explosion
Agentic coding tools didn't just make developers faster. They changed the size of the app store.
- New App Store releases roughly doubled to about 560,000 in the first half of 2026, nearly matching the ~600,000 apps published across all of 2025.[1] Appfigures sees the same curve from a different dataset: worldwide new releases up 60% year-over-year in Q1 2026, iOS up 80%.[1]
- There are now roughly 3.3 million unique non-gaming apps from 1.25 million publisher accounts across Google Play and the App Store.[2]
- RevenueCat counts almost 15,000 new subscription apps launching every month, up from about 2,000 three years ago.
Sensor Tower's read on the same period, delivered on stage at the TikTok Digital Summit in Istanbul, was blunt: downloads across most genres have levelled out, the market is saturated, and creative diversification is now essential.[6]
~560,000 new App Store releases in H1 2026 — roughly double
+2% download growth over the same six months, to 17.6 billion
Here is the part that matters for anyone doing marketing. Over the same six months in which releases doubled, downloads grew 2%, to 17.6 billion.[1] Supply went vertical. Demand stayed on the same gentle slope it has been on for years.
Attention didn't get the memo
Twice as many apps are competing for the same feed, and the feed itself hasn't grown. Neither has the human on the other side of it.
1.5 seconds of attention is enough for a person to encode an ad. 85% of digital ads never get 2.5.[4]
That's the whole game in two numbers. You're not competing on features or on price at the moment of the scroll. You're competing for a second and a half, against twice as many apps as last year, in a feed where everything is designed to keep moving.
The old answer was to buy more of that attention. That answer is getting more expensive: marketing budgets grew 7.7% in 2025, which is flat once you account for inflation.[8] Meanwhile TikTok's own forecast is that content demand will grow 5× by 2027, because culture now updates daily and creative demand has entered what they call an always-on cycle.[3]
So the constraint isn't budget and it isn't targeting. It's how many distinct, relevant things you can put in front of people, and how fast you can replace the ones that stop working.
The velocity gap: what the leaders are already doing
The interesting thing about Sensor Tower's creative data is that the shift has already happened at the top. The teams winning downloads in non-gaming categories restructured their creative production two years ago.
| Category (US) | New unique creatives per advertiser per month, Q1 2026 vs. Q1 2023 |
|---|---|
| Health & Wellness | +94% |
| Financial Services | +108% |
| Lifestyle | +145% |
| Job & Education | +173% |
Source: Sensor Tower, presented at TikTok Digital Summit 2026, Istanbul.
Volume went up, hard. The top 50 Health & Wellness apps in the US averaged more than 450 distinct ad creatives in 2025, up 41% on 2023.[6]
Lifespans collapsed. In financial services, roughly half of all creatives are now retired within 10 days of launch, up from about a quarter in early 2023. Creatives that last 60+ days have dropped from more than half the mix to around a fifth.[6] A creative is no longer an asset. It's a test with a short shelf life.
Video overtook image. As AI video quality improved, the average advertiser in Financial Services, Health & Wellness, and Jobs & Education flipped from image-heavy to video-heavy creative mixes between 2023 and 2026.[6] The video line crosses the image line in every one of those categories.
And it correlates with growth better than money does. Among the top 20 US finance apps, growth in creative count correlates with download growth at 0.474, ahead of growth in ad impressions (0.410) and growth in ad spend (0.337).[6] Those top 20 apps added an average of 346 distinct creatives year-over-year; the leaders added more than 1,000. Klarna, one of the examples Sensor Tower highlighted, paired new creative testing with 37% year-over-year download growth.[6]
| Metric growth (top 20 US finance apps, Q2 2025–Q1 2026) | Correlation with download growth |
|---|---|
| Number of creatives (+346 avg.) | 0.474 |
| Ad impressions (+178M avg.) | 0.410 |
| Ad spend (+$30M avg.) | 0.337 |
Source: Sensor Tower, presented at TikTok Digital Summit 2026, Istanbul.
Gaming, which usually runs two years ahead of non-gaming on UA tooling, shows where this goes next. AppsFlyer's State of Gaming 2026 has the top UA spenders shipping 2,400–2,600 creatives per quarter per app, up 25–30% year-over-year on the back of AI production. The mid tier, apps spending $1–4M a quarter, is going backwards at around 1,000: big enough that iteration got slow, not big enough to fund industrial-scale production.[7] That squeeze is exactly what's arriving in non-gaming now.
84% of marketers already say they lean on generative AI to keep up.[8] The question is no longer whether to use it, but whether your pipeline is built around it or just sprinkled with it.
The platforms are rewriting the rules in parallel
This is the part most teams miss. The ad platforms aren't passive here. Both TikTok and Meta have rebuilt their systems around the assumption that advertisers will bring far more creative than they used to, and both now reward the teams that do.
TikTok: "the old creative model was built for control"
TikTok's pitch at the Istanbul summit was unusually direct about the problem. The traditional pipeline (brief, produce, approve, launch, evaluate) was designed to protect quality, and it does. What it can't do is keep pace with culture. Budgets, talent, and time are finite; manual handoffs across teams slow momentum; slow pipelines turn creative production into a cost center.[3]
Their diagnosis of where the industry is right now: the current model solved volume, but not relevance. More creative, most of it noise. And their stated risk: scaling with low-affinity assets weakens brand equity.[3]
TikTok's answer is to make creative production itself agentic. Symphony, their generative creative suite, is being repositioned from a set of tools into infrastructure, with a Creative Studio, an AI search layer over existing brand content, a creator-matching search, and a Symphony Agent that pulls cultural signals, trends, and top-performing patterns straight into the production loop. Their summary line: know what to make, create it faster, scale it through infrastructure.[3]
Read between the lines and the message to advertisers is clear. TikTok expects the number of creatives per advertiser to keep multiplying, and it does not believe that can be done by hand.
Meta: Andromeda made creative the targeting
Meta went a level deeper and changed how ads get chosen in the first place.
Andromeda is Meta's retrieval engine, the first stage of ad delivery where the system narrows a pool of tens of millions of ads down to the few thousand it will consider showing a given person. Meta's engineering team reports it enabled a 10,000× increase in the complexity of the models used at that stage, along with a 6% improvement in recall and 8% improvement in ad quality on selected segments.[5] It rolled out from late 2024 and reached every account by October 2025.
What changed for advertisers is the direction of the question. Meta's own framing is that with AI-enabled tools, the focus has shifted from niche targeting to creative diversification as the best lever for finding the most relevant audiences. The system no longer asks which person fits this ad. It reads the creative itself (visual elements, opening seconds, copy structure, format) and asks which ad fits this person.[5]
Two practical consequences follow.
First, diversity is the signal. Andromeda learns from variety, not repetition. Five cosmetic variations of the same ad teach it almost nothing about who else you could reach. Twenty conceptually different ads (different hooks, personas, use cases, formats) give it twenty different ways to find your next user. Practitioner analyses since the rollout consistently find that near-identical creatives get treated as one entity and end up competing with each other rather than expanding reach.[5]
Second, fatigue is faster. When the system can match a creative to exactly the people it fits, it exhausts that pool quicker. The 10-day lifespans in Sensor Tower's finance data aren't a coincidence. They're what a retrieval system this efficient does to a creative.
Put the two platforms together and the strategy is the same on both: bring many more creatives, make them different from each other, and replace them on a weekly cadence rather than a quarterly one.
What this means for a non-gaming app team
Here's how it lands by role.
If you run performance marketing: your unit of work is no longer the campaign, it's the concept. Plan for a weekly rhythm where every winning ad has a live challenger. Track hook rate and CPI, but also install-to-trial and install-to-paid per creative, because Andromeda will happily find you cheap installs that never convert. Kill on the second set of numbers, not the first.
If you're a motion designer: the job is shifting from finishing one polished hero video to directing a system that produces thirty defensible variations of it. Your taste is the scarce input; the rendering isn't. The teams pulling ahead treat a reference ad as a brief, not a deliverable.
If you're on product or data: creative is now upstream of your funnel, not just next to it. The ad that brings a user in sets their expectation for the first screen. In the apps we work with, carrying the creative's design and promise through a deep link into onboarding lifts install-to-subscription by 30% on average and up to 120% in the best-aligned funnels. Creative velocity without onboarding continuity leaves that on the table.
If you're an indie developer: this is the one place the market tilted in your favor. Agentic coding made your competitors cheap to build. AI creative production makes a 20-concept-a-week testing cadence possible for a one-person team. The studios stuck at 1,000 creatives a quarter are the ones with a legacy pipeline to protect. You don't have one.
Where Pantheon fits
Pantheon exists because we hit this wall ourselves. Building the app was the easy half. Feeding two ad platforms that now expect dozens of distinct concepts a week, in every market we sold in, was the half that broke.
So we built the pipeline the platforms are asking for:
- Ad Intelligence (Sable) shows you what's actually working in your category right now, so you know what to make before you make it.
- Ad Create & Clone (Serafine) takes any ad you admire as a reference and generates on-brand ads for your app from it. Reference-to-ad, not templates, not blank-canvas prompting.
- Variations & Localisations (Echo) turns each concept into the OS-specific, market-specific, format-specific cuts that Andromeda and TikTok's retrieval systems reward, in every language you sell in.
- Campaign Management (Cormac) launches them, watches the numbers that matter, and rotates in challengers before the winners fatigue.
The whole loop, from reference ad to live campaign, runs in hours instead of weeks. That's the difference between a creative pipeline built for control and one built for the always-on cycle.
Bring one reference ad. See what a week of creative looks like when the volume problem is solved.
Sources
- Sensor Tower, H1 2026. New App Store releases roughly doubled to ~560,000 in the first half of 2026, approaching the ~600,000 published across all of 2025; downloads over the same period grew 2% to 17.6 billion. Corroborated by Appfigures: worldwide new releases +60% YoY in Q1 2026, iOS +80%.
- 42matters live store scan, 23 August 2026. Google Play 2,538,643 listings (88.3% non-gaming); Apple App Store 2,550,196 (90.4% non-gaming); 37% cross-listing rate. Deduplicated: ~3.3M unique non-gaming apps, ~1.25M publisher accounts.
- TikTok Digital Summit 2026, Istanbul. "Creative demand has entered an always-on cycle." Content demand expected to grow 5× by 2027.
- VCCP Media × Amplified, 2025.
- Meta Engineering, "Meta Andromeda: Supercharging Advantage+ automation with the next-gen personalized ads retrieval engine," December 2024; Meta advertiser communications, March–April 2025.
- Sensor Tower, presented at TikTok Digital Summit 2026, Istanbul (creative count and lifespan data, US, Q1 2023–Q1 2026).
- AppsFlyer, State of Gaming 2026.
- Adobe, 2025 (marketing budgets); Gartner CMO Spend Survey (GenAI adoption).