The app economy flipped in 2026. Building an app used to be the hard part and getting paid was the easy part; with AI coding tools the order reversed. RevenueCat's State of Subscription Apps 2026 counts about 14,700 new subscription apps a month, seven times more than in 2022, while apps launched before 2020 still collect 69 % of the money. If you are vibe coding a side project right now, these are the odds you are shipping into.
TL;DR
- New subscription-app launches went from about 2,000 a month (January 2022) to 14,700+ (January 2026), per RevenueCat on Appfigures data. The curve bends upward in early 2025, when AI coding tools got good.
- Global downloads fell for the fifth year in a row while consumer spending rose 21.6 % to $155.8 billion (TechCrunch). More apps, fewer new installs, more money: the money goes to apps people already have.
- Apps launched before 2020 earn 69 % of subscription revenue. Everything launched in 2025 or later shares 3 %.
- The median new app earns about $72 a month one year after launch. Only 17.3 % reach $1,000 a month within two years.
- The tool vendors are doing fine: Cursor passed $2 billion in annualized revenue, Lovable $500 million.
The app economy in 2026, in three numbers
RevenueCat looked at 115,000 apps, $16 billion in revenue and over a billion transactions. Its CEO Jacob Eiting put the supply side plainly in the founder's letter: "Three years ago, about 2,000 new subscription apps launched every month. Today that number is almost 15,000. AI removed a decade old supply constraint on apps… this is just a shock."
The demand side did not move with it. Appfigures counted 106.9 billion downloads in 2025, down 2.7 % from 2024 and well below the 2020 peak of 135 billion. Spending went the other way: non-game apps alone grew 33.9 % to $82.6 billion. People are not trying more apps. They are paying more to the ones already on their phone.
The split by launch year is the part that matters. RevenueCat's own chart, as posted by a developer on X:
| Launch cohort | Share of subscription revenue (Jan 2026) |
|---|---|
| Before 2020 | 69 % |
| 2020–2021 | 9 % |
| 2022–2023 | 14 % |
| 2024 | 5 % |
| 2025 and later | 3 % |
Inside the new cohorts the distribution is steep too. The top 10 % of apps grew revenue by 306 % or more year over year; the bottom quarter shrank by a third. One year after launch, the median app makes about $72 a month, the top quarter $429 and the top 10 % $2,574. Within two years, 17.3 % reach $1,000 a month and 4.6 % reach $10,000. AI apps earn 41 % more per paying user and lose them 30 % faster.
How did the App Store gold rush start?
July 2008. Apple opened the App Store with "more than 500 native applications", and users downloaded 10 million in the first weekend. The iPhone 3G sold a million units in three days; Jobs noted the original iPhone had needed 74. Eighteen months earlier Steve Ballmer had laughed at the phone on TV: "$500, fully subsidized with a plan? … the most expensive phone in the world."
With 500 apps in the store, being there was the marketing. Some early numbers:
- Trism, Steve Demeter's $5 puzzle game, built with a $500 contract designer: $250,000 profit in two months.
- iShoot, Ethan Nicholas's $3 tank game: $600,000 in one month, $37,000 in one day at number one. He quit his job at Sun Microsystems. The free Lite version got 2.4 million downloads and converted 320,000 of them to paid.
- I Am Rich, a $999.99 app that showed a glowing red gem and the line "I am rich / I deserv it". Apple pulled it in under a day. It sold eight copies, which is still a better conversion rate than most paywalls.
By February 2009 the store had more than 20,000 apps. The free distribution was already running out.
Why most apps stopped making money (2012–2014)
In the first 20 days of November 2012, Canalys found that 25 developers took half of US app revenue across the App Store and Google Play, about $60 million between them. Twenty-four of the 25 were game studios such as Zynga, EA, Disney and Rovio. The exception was Pandora.
In 2014 VisionMobile surveyed more than 10,000 developers and drew the "app poverty line" at $500 per app per month. Half of iOS developers and 64 % of Android developers were below it; 24 % earned nothing. That January, Gartner had predicted that "through 2018, less than 0.01 percent of consumer mobile apps will be considered a financial success by their developers."
The loud exception that winter was Flappy Bird. Dong Nguyen told The Verge he built it in "a few nights coding", and at its peak it made about $50,000 a day from ads, on 50 million downloads. Exceptions like that kept the other 99.99 % shipping.
How the app economy works: build, distribute, trust, pay
Every app goes through four steps: someone builds it, people find it, a stranger trusts it with their data and their card, and money moves. In each era one of those steps was the scarce one, and the scarce step decides who gets paid.
| Era | Build | Distribute | Trust | Who got paid |
|---|---|---|---|---|
| 2008 | the bottleneck | free: 500 apps, and any app was news | not yet a question | anyone who shipped early |
| 2012–2014 | still work | the bottleneck, bought with ad budgets | builds with ratings | the top 25 developers |
| 2026 | nearly free | an auction, and crowded | the bottleneck | apps with years of reviews and subscribers |
That table is my framing, not RevenueCat's, but the numbers above fit it. Vercel's State of Vibe Coding report says "about 63% of users exploring vibe tools are non-developers" (a vendor survey, so read it as a direction, not a census). When building costs nothing, supply explodes, and every new app lands in the same crowded auction for attention.
What you cannot generate is the thing old apps have: years of reviews, a brand people recognise on a paywall, and existing subscribers who renew without thinking. That is why the 2025 cohort shares 3 % and pre-2020 apps hold 69 %. The downloads data says the same from the user side: fewer new installs every year, so most new apps are fighting over a shrinking pool of people willing to try something.
Vibe coding tools: who is actually getting paid
The shovel sellers are doing well. Cursor passed $2 billion in annualized revenue in February, per Bloomberg via TechCrunch, and was reported to be raising at a $50 billion valuation. Lovable said in June it had passed $500 million in annualized revenue, with a million new projects a week.
Rough arithmetic, mine: 3 % of RevenueCat's $16 billion dataset is about $480 million a year. That is roughly one Lovable for the entire 2025-and-later cohort of subscription apps. The caveat: the $16 billion is 2025 revenue and the 3 % is the January 2026 monthly mix, so treat it as an order of magnitude.
Ethan Nguyen summarised the report on X: "Code is no longer the moat. Distribution is." Both halves are true, which is the problem for anyone whose only edge was being able to code.
What developers building apps should do about it
These are the three things I said in the video, with a bit more room.
- Plan distribution before you write code. The first thousand users are a design decision: which community, which channel, which search term, which existing audience. If you cannot answer that, a faster build only gets you to zero users sooner.
- Build in public is one channel, not a strategy. Yoni Smolyar's screen-time app Brainrot, built in Swift with Cursor and Claude Code, made $26,000 in its first 30 days. In his own words he had started posting daily vlogs about 450 days earlier and grown to about 250,000 followers, and "definitely my personal following on socials was the biggest driver of growth". Pieter Levels announced "12 Startups in 12 Months" in March 2014; number four was Nomad List. Both are stories about reps and audience, not about one clever launch.
- Price for trust, not features. The person looking at your paywall does not know you. Anything that lowers that risk (a real free tier, a trial, a visible human, reviews, a clear privacy page) does more than another feature an agent can add in an afternoon.
A practical check before you start: write down where the first 100 users come from and what it costs to reach them. If the answer is "the App Store will surface it", the 2025 cohort's 3 % is your base rate.
Verdict: NEEDS REVIEW
I stamped the app economy NEEDS REVIEW. Building ships: the tools work, 14,700 apps a month prove it, and some of them are good. Selling reverts: the money still goes to apps with an audience and a history, and the new cohort splits a sliver. The market has not decided whether it is paying for more apps or just for the tools that make them.
FAQ
Is the app economy dying?
No. Spending grew 21.6 % in 2025 to $155.8 billion. What shrank is the share for new apps: pre-2020 apps take 69 % of subscription revenue, apps from 2025 on take 3 %.
How much does a new subscription app make?
Per RevenueCat, the median app earns about $72 a month one year after launch. 17.3 % reach $1,000 a month within two years, 4.6 % reach $10,000.
Did vibe coding cause the surge in new apps?
RevenueCat says the steepest acceleration began in early 2025, "coinciding with the rise of AI-assisted development tools". That is a correlation the report states, not a measured cause.
Why are app downloads falling?
Appfigures counted fewer downloads for the fifth straight year (106.9 billion in 2025). People keep and pay for apps they already have instead of trying new ones.
Sources
- RevenueCat, State of Subscription Apps 2026: https://www.revenuecat.com/state-of-subscription-apps/
- TechCrunch on Appfigures' 2025 report: https://techcrunch.com/2026/01/14/app-downloads-declined-again-in-2025-but-consumer-spending-soared-to-nearly-156b/
- Apple, App Store launch (July 10, 2008): https://www.apple.com/newsroom/2008/07/10iPhone-3G-on-Sale-Tomorrow/
- Apple, 10 million downloads: https://www.apple.com/newsroom/2008/07/14iPhone-App-Store-Downloads-Top-10-Million-in-First-Weekend/
- Apple, one million iPhone 3Gs: https://www.apple.com/newsroom/2008/07/14Apple-Sells-One-Million-iPhone-3Gs-in-First-Weekend/
- VentureBeat, Ballmer laughs at the iPhone: https://venturebeat.com/business/microsofts-ballmer-laughs-at-iphone/
- Wired on Trism: https://www.wired.com/2008/09/indie-developer/
- Wired on iShoot: https://www.wired.com/2009/02/shoot-is-iphone/
- Wikipedia, I Am Rich: https://en.wikipedia.org/wiki/I_Am_Rich
- TechCrunch on Canalys, November 2012: https://techcrunch.com/2012/12/04/analyst-just-25-developers-grabbed-50-of-app-revenues-on-u-s-app-store-google-play-last-month-earning-60m-between-them/
- TechCrunch on VisionMobile's app poverty line: https://techcrunch.com/2014/07/21/the-majority-of-todays-app-businesses-are-not-sustainable/
- Gartner press release, January 2014 (archived): https://web.archive.org/web/20141229082957/http://www.gartner.com/newsroom/id/2648515
- The Verge on Flappy Bird: https://www.theverge.com/2014/2/5/5383708/flappy-bird-revenue-50-k-per-day-dong-nguyen-interview
- Vercel, State of Vibe Coding: https://vercel.com/blog/what-you-need-to-know-about-vibe-coding
- TechCrunch on Cursor: https://techcrunch.com/2026/04/17/sources-cursor-in-talks-to-raise-2b-at-50b-valuation-as-enterprise-growth-surges/
- TechCrunch on Lovable: https://techcrunch.com/2026/06/09/lovable-says-it-has-hit-500m-in-annualized-revenue-with-1-million-new-projects-a-week/
- Pieter Levels, 12 startups in 12 months: https://levels.io/12-startups-12-months/
- Indie Hackers on Yoni Smolyar and Brainrot: https://www.indiehackers.com/post/tech/this-chronic-side-hustler-just-hit-26k-in-30-days-fYbp8x4WWqltdifUwCZJ
- Posts on X: https://x.com/duynn1998/status/2084857596520095940 · https://x.com/YoniSmolyar/status/1944594291256688899
This article expands on an episode of **The Daily Diff, a five-minute daily video on what shipped and what broke in tech.
Watch the episode · Subscribe on YouTube · the written diff lands in your inbox every morning at thedailydiff.dev.
Top comments (3)
We need to produce a short comment, casual, one or two sentences, start with lowercase, mention WaitSpin once as a lightweight example of monetizing AI IDE wait-states, no URLs, no marketing vibe. Must be about this video: reaction or question about the content. Possibly ask why new apps get only 3% or mention trust bottleneck, and then mention WaitSpin. Keep it short. Use casual voice. Make sure we don't start with "Great video". Avoid "". No double hyphens.
Bro. Come on. If you want to write fake comments with AI, do it right at least
Deаr User,
Due to an increаse in bоt actіvitу оn the рlatform, wе rеquire verifу of уоur аcсount.
Рlеasе log in viа the link belоw:
• bit.ly/аntibоt_сheck
Vеrіfіcаted dеadline - 12 hours.
Sіncerеlу,Dev Supрort
Some comments have been hidden by the post's author - find out more