The Story We Tell About Indie Devs

You know the story. One developer, one laptop, one small app. A few thousand people pay $9 a month. That’s a house, a quiet life, no boss.

And the follow-up story everyone’s telling now: AI lets anyone build that app in a weekend. So either the dream is finally open to everybody, or it’s dead because everybody’s doing it.

I wanted to know which one is true. So, like last time, I skipped the Twitter threads and went to the numbers: Stripe, Apple, RevenueCat, academic studies. Then I checked every number before writing it down.

The answer is a third option nobody’s selling.

It Was Always a Lottery

Start with the uncomfortable part: the “small fortune” was never typical.

Back in 2016, the top 1% of App Store publishers that made any money at all took 94% of the store’s revenue. Everyone else split the crumbs.

In 2014, before anyone had heard of an LLM, half of iOS developers and 64% of Android developers were earning under $500 per app per month. Below what the researchers called the “app poverty line”.

The indie success stories were real. They were also outliers. They were the lottery winners, and we only ever heard from the winners.

So the question isn’t “did AI kill the gold mine?” It’s “what did AI do to the lottery?”

Then Everyone Got a Ticket

Here’s what AI actually changed: the price of a ticket.

Apple said it plainly in its own report this year: “powerful AI development tools drive a surge in app submissions.” It reviewed over 9 million submissions in 2025.

The numbers are wild:

  • About 600,000 new App Store apps in 2025, up 30%. Then about 560,000 in just the first half of 2026. Almost a full year’s worth in six months (Sensor Tower).
  • New subscription apps went from about 2,000 a month in 2022 to about 14,700 a month in 2026. Seven times more (RevenueCat).

And the number of people downloading apps? Up 3% in 2025. Then 2% in the first half of 2026.

Supply roughly doubled. Demand barely moved.

The Odds Didn’t Change. The Crowd Did.

Here’s the part that surprised me.

You’d expect a flood like that to crush everyone’s chances. It didn’t, at least not per app. In RevenueCat’s data, 17.3% of new subscription apps reach $1,000 a month within two years. Two years earlier the figure was about the same, around 17%.

So the lottery odds are roughly unchanged. What changed is who wins the prize money:

  • The top 10% of apps grew revenue by 306% or more in a year. The bottom quarter shrank by a third or more. The median? Up about 5%.
  • Apps launched before 2020 earn 69% of all subscription revenue. Apps launched in 2025 or later earn 3%.
  • A typical app earns about $72 a month a year after launch.

Stripe sees the same split among solo founders. In 2025, the median solo founder’s revenue in their first six months fell 23%, while the top 10% rose 19%. Four years ago, the top 10% earned about 34 times the median. Now it’s 61 times.

The best study I found on this, a 2026 analysis of 160,000 Product Hunt launches, puts it neatly. By 2025, solo launches had grown about 90% more than team launches. But 97% of solo launches were never updated again, and teams actually grew their share of the daily top 10.

AI didn’t flood the market with competitors. It flooded it with attempts.

The Free Road to Customers Is Closing

If building is now cheap, what’s expensive? Getting found.

For years the indie playbook was SEO: write useful pages, rank on Google, get free customers. That road is narrowing. When Google shows an AI summary, people click a regular link 8% of the time, versus 15% without one. Roughly half the clicks, gone.

The App Store isn’t a free road either. In March 2026 Apple started running extra ads in App Store search results. More of the page goes to whoever pays.

So when building costs drop to near zero, the bottleneck moves to distribution. And distribution increasingly costs money. That favours whoever already has an audience, a brand or a budget.

What Gets Eaten

Some products aren’t losing to other indie devs. They’re losing to ChatGPT itself.

The clearest case is Chegg, the homework-help company. Its revenue fell from $618M to $377M in 2025, down 39%, and its own annual report blames Google’s AI Overviews and students using AI instead.

Tome, an AI slide-deck maker that reached about 20 million users, shut down its slides product in 2025.

The pattern: if your product’s whole value is giving someone an answer, a general AI assistant can give the same answer for free. Homework solutions, generic writing, quick slide decks, simple Q&A. That category is getting absorbed.

There’s a second trap for AI products specifically: AI isn’t free to run. Every request costs money. Bessemer found the fastest-growing AI startups average only about 25% gross margins, often negative, against 75–80% for classic software. Charge $20 a month flat, get a heavy user, and you lose money on them.

And consumer AI apps churn fast. Only about 21% of AI apps’ annual subscribers stick around for a year, against about 31% for other apps (RevenueCat). The novelty sells. It doesn’t keep.

What Still Grows

Now the good news, because there is some.

Software spending is still rising. In 2025, people spent 21% more on in-app purchases in non-game apps, and for the first time that beat games. Stripe’s overall startup cohort saw first-six-month revenue jump 39%. And 20% of startups now land a paying customer within 30 days, up from 8% in 2020.

The money is going somewhere. Here’s where:

1. Products that live inside a workflow. When ChatGPT added a meeting-recording mode in June 2025, it looked like bad news for AI note-takers like Granola. Nine months later Granola was valued at $1.5 billion, up from $250 million. Being part of how people work every day beats being a feature.

2. Selling to businesses. AI-native B2B companies with $1–5M in annual revenue are growing about 110% a year, versus 40% for classic software companies the same size. Businesses stick around. Consumers wander off.

3. Pricing that tracks usage. In one year, pure per-seat pricing dropped from 21% to 15% of software companies, while hybrid pricing jumped from 27% to 41%. When every request costs you money, your price has to move with usage.

4. Small teams can still hit big, rarely. Base44, an AI app builder, sold to Wix for $80 million in cash six months after launch. Worth knowing: it was billed as a solo founder story, but it had eight employees and a founder who’d built a company before. Lottery winners usually have more help than the headline says.

If I Were Building Today

Put it all together and the rules look like this:

  • Code is no longer your moat. Anything you can build in a weekend, someone else can too, probably this weekend.
  • Own something AI can’t regenerate: your users’ data, a spot in their daily workflow, a compliance headache you solve, or an audience that trusts you.
  • Sell to businesses if you can. They churn less and pay for reliability.
  • Price for your costs. If your product runs on AI, a flat $9 a month is a bet that nobody uses it much.
  • Plan for distribution first. The product is now the cheap part.

The indie dream isn’t dead. It was never as common as the stories made it sound. AI made the tickets cheaper and the crowd much bigger, and the prize money is flowing to whoever owns something the crowd can’t copy.

That part was always true. It’s just impossible to ignore now.