Tony Wang13 min readMore Apps, Fewer Hits: 18,000 App-Store Winners, by Birth Year
AI coding flooded the app stores — 2025 had the most new apps since 2016. But of ~18,000 apps that hit traction and survive today, the birth-year peak is 2019.
The pitch for AI coding was that it would unleash a wave of new software: anyone with an idea could "vibe code" an app in a weekend. On the supply side, it worked. Apple just logged its biggest release year since 2016 — roughly 557,000 new App Store apps in 2025, up 24% — and early 2026 is running even hotter, with iOS releases up around 80% year-over-year. Appfigures, which tracks this, puts "AI and Vibe Coding" at the top of its list of reasons.
So we asked the obvious follow-up question the headlines skip: of all the apps ever made, which ones actually made it — and when were they born? We pulled every app in the App Store and Google Play that has crossed a real traction bar and is still live today, and bucketed each one by its store release date.
The answer is a paradox. Of about 18,000 apps that reached traction and survive in mid-2026, the biggest single birth-year is 2019 — and every year since has produced fewer lasting hits, even as the number of apps created climbed to a decade high. More apps than ever are being born. Fewer than ever are being found.
What this means
The one-line version: supply is booming and success is concentrating. What to do about it depends on who you are:
- App builders: 'I can build it now' is no longer the constraint — everyone can. The 2025 traction cohort is the thinnest in a decade, and two-thirds of it is games. If you're shipping a utility or AI tool, budget for distribution and ASO like it's the whole job, because it now is.
- Investors & analysts: read release counts as supply and traction cohorts as demand. Appfigures' record 2025 release year and our shrinking survivor cohorts are both true and not in tension — creation is up, breakthrough rate is down. Judge apps on ratings, installs and rank, not on how many shipped.
- Platform & policy watchers: Apple and Google spent 2022–2025 purging and gatekeeping (Google Play's live catalog fell ~47% in a year). The stores got harder to survive on at the same moment they got easier to publish to — that scissor is the story.
- Data & API buyers: the durable signal is release dates, rating counts and chart ranks pulled as JSON and re-pulled on a schedule. A single snapshot like this decays as cohorts age and the charts turn over.
The creation boom is real
Start with the thing everyone agrees on. New-app releases on the App Store collapsed from their 2016 peak, bottomed out in 2023, and have since come roaring back:
The timing lines up with the tools. GitHub Copilot went generally available in June 2022; ChatGPT launched that November; Cursor and Claude arrived in early 2023; and Andrej Karpathy coined "vibe coding" in February 2025 — right as the release curve turned sharply upward. Google Play tells a messier version of the same story: its live catalog shrank about 47% in a year (roughly 3.4M to 1.8M) under waves of policy and quality purges, yet its new releases were still ticking up. On both stores, more apps are being made.
But the winners were born earlier
Now the same question asked of the apps that worked. Here is every one of our ~18,000 traction-survivors, counted by the year it was first released. Because the two censuses are nearly the same size (~8.8k iOS, ~9.2k Android), the raw counts are directly comparable — and both tell the same story:
The shape is unambiguous. Traction-winning apps accumulate on a plateau from 2016 through 2020, peak in 2019 on both stores (788 iOS, 806 Google Play), and then fall every single year. The AI era — everything from 2022 on — sits entirely on the downslope.
Show the full year-by-year table (both stores)
| Release year | iOS (≥5k ratings) | Google Play (≥120k) | Combined | Combined share |
|---|---|---|---|---|
| 2008 | 12 | — | 12 | 0.1% |
| 2009 | 117 | — | 117 | 0.7% |
| 2010 | 249 | 254 | 503 | 2.8% |
| 2011 | 339 | 382 | 721 | 4.0% |
| 2012 | 390 | 541 | 931 | 5.2% |
| 2013 | 470 | 604 | 1,074 | 6.0% |
| 2014 | 481 | 800 | 1,281 | 7.1% |
| 2015 | 492 | 767 | 1,259 | 7.0% |
| 2016 | 706 | 774 | 1,480 | 8.2% |
| 2017 | 692 | 804 | 1,496 | 8.3% |
| 2018 | 697 | 800 | 1,497 | 8.3% |
| 2019 | 788 | 806 | 1,594 | 8.9% |
| 2020 | 765 | 752 | 1,517 | 8.4% |
| 2021 | 672 | 599 | 1,271 | 7.1% |
| 2022 | 574 | 426 | 1,000 | 5.6% |
| 2023 | 517 | 372 | 889 | 5.0% |
| 2024 | 479 | 305 | 784 | 4.4% |
| 2025 | 299 | 168 | 467 | 2.6% |
| 2026 (H1) | 45 | 18 | 63 | 0.4% |
82% were born before ChatGPT
Fold the years into eras and the concentration is stark. Split the combined census at the arrival of modern AI coding:
Eighty-two percent of the apps carrying real usage today were shipped before ChatGPT even launched. The three-plus years of the AI-coding boom — 2022 through mid-2026 — have so far contributed 18% of the surviving winners, and the "vibe coding" window (2025 on) just 3%. On iOS the AI era looks a little healthier (22% of winners) than on Google Play (14%), consistent with Apple's release rebound versus Google's purge-driven contraction.
"But those apps are just too young"
This is the honest objection, and it's partly right. An app released in 2025 has had at most eighteen months to climb to 5,000 or 120,000 ratings; some of today's thin recent cohorts will fatten as they age. That right-censoring is real, and it's why we don't make much of the 2025–2026 bars on their own.
But censoring can't explain the whole decline, because the fall starts in cohorts that are fully aged:
The 2021 class of survivors is already smaller than 2016's, and 2022's is smaller than 2021's — and all of these apps are well past the age where traction is established. You can also attack the censoring worry from the other side and lower the bar: drop the iOS threshold five-fold, to any app with 1,000+ ratings (19,305 apps), and the recent cohorts do climb back into view — yet even then 74% of survivors were first released in 2021 or earlier. However you set the traction floor, the AI era is a minority of the winners. The breakthrough rate started declining around 2020–2021 and the AI-coding surge did not reverse it. If anything, the extra supply made discovery harder.
What still breaks through: games
One more cut explains where the surviving new apps come from. Break the iOS traction-winners into categories and compare the recent cohort to the old one:
of iOS traction-winners released in 2025 or later
of iOS traction-winners from the pre-AI era
Games still break through because gaming has a working discovery machine — top-charts, virality, ad networks, and a rating cadence that surfaces hits fast. The categories that AI coding most plausibly lowered the barrier to — utilities, productivity, "AI assistant" tools — are exactly the ones missing from the recent winners' list. That squares with our two earlier studies: the App Store's AI-branded catalog exploded to 8,505 apps with a median of 10 ratings. Vibe coding fills the catalog; it doesn't fill the charts.
The bottom line
Two facts, both true: app creation is at a decade high, and the rate at which new apps reach real users has been falling since 2019. AI coding is a credible, well-timed cause of the first — Appfigures and Apple both say so — and it is simply orthogonal to the second. Building was never the bottleneck that decided which apps mattered; distribution and store gatekeeping were, and AI touched neither. The result is a market where it has never been easier to ship an app and never been harder for one to be found.
How we did this (and the caveats)
We enumerated two tiers of the Crawlora apps dataset in full — every iOS app with ≥5,000 ratings (8,784) and every Google Play app with ≥120,000 ratings (9,172 with a parseable release date, out of 9,753) — through a structured app-store API, US storefront, June 2026, and grouped each app by its store release date. The thresholds are deliberately different because Google Play rating counts run an order of magnitude higher than iOS; they select comparable traction, not equal numbers, and the shape of the curve is the same whether you read the stores together or apart.
The caveats, plainly. Survivorship: this is a snapshot of apps alive in June 2026, so any app that was delisted, abandoned or purged is absent — which undercounts old cohorts and means these counts are a floor, never a census of everything ever built. For total creation we defer to Appfigures' release series, which we cite rather than reproduce. Right-censoring: the newest cohorts are young and will grow, so we lean on the mature 2013–2022 years for any claim about decline and treat 2025–2026 as directional. Traction is a scale proxy, not a quality verdict — a small recent cohort means few apps found an audience yet, not that the rest are bad. And rating counts aren't installs — Apple doesn't publish installs, and Google's are bucketed. Every figure here is reproducible from the dataset, and the crawl behind it was a few hundred API calls.
Read the App Store and 40 other platforms as clean JSON
Crawlora turns app stores, search engines, marketplaces, and social platforms into normalized JSON — handling proxies, rendering, and anti-bot — and bills pay-on-success. 2,000 free credits a month, no card.
Frequently asked questions
Did AI and 'vibe coding' cause a surge of new apps?
On the supply side, yes. Apple logged its biggest release year since 2016 in 2025 — about 557,000 new App Store apps, up 24% — and early-2026 iOS releases are running roughly 80% ahead of last year, with Appfigures naming 'AI and vibe coding' as a leading cause. But it did not translate into more successful apps: of ~18,000 apps that reached real traction (5,000+ iOS ratings or 120,000+ Google Play ratings) and are still live in June 2026, the single biggest birth-year cohort is 2019, and 82% were first released in 2021 or earlier — before ChatGPT existed.
When were the app stores' most successful apps actually created?
The traction-winning apps that survive today accumulate on a plateau from 2016 to 2020, peak in 2019 (788 on iOS, 806 on Google Play), and then decline every single year after. The entire AI era (apps born 2022 or later) accounts for just 15% of surviving winners, and the 'vibe coding' window (2025 on) only 3%.
Aren't recent apps just too young to have reached traction yet?
Partly — that right-censoring is real, which is why we don't lean on the 2025–2026 cohorts. But it can't explain the whole decline, because it starts in fully-aged cohorts: the 2021 class of survivors (1,271 combined) is already smaller than 2016's (1,480), and 2022's (1,000) is smaller still, with four to five years to prove themselves. The breakthrough rate began falling around 2020–2021 and AI coding did not reverse it.
What kinds of new apps still break through?
Increasingly, games. 67% of iOS traction-winners released in 2025 or later are games, versus 26% among apps from before 2022. Gaming still has a working discovery machine — top charts, virality, ad networks — while the utility, productivity and 'AI assistant' categories that AI coding most lowered the barrier to are largely missing from the recent winners' list.
How many apps are on the App Store and Google Play?
Crawlora's mobile-app dataset indexes about 4.1 million apps across the two stores (roughly 1.17M on the iOS App Store and 2.94M on Google Play) as of June 2026. Only about 18,000 of them have crossed the traction bars used in this study — a reminder that the vast majority of apps, old and new, are never found by anyone.