Everyone Has Claude Now: Where the Moat Moved

Building software got close to free. Selling it did not.

A thread on Israeli tech X this week made the point bluntly. Its author keeps meeting people, often unemployed or mid-career-switch, who vibe-coded a project and talk about it as their future livelihood. His advice: zoom out. Everyone has Claude, and nobody is going to line up when you launch.

He applied it to himself too. A product he started four years ago can now be built in a day with Opus 5.5. Even if the copies are worse, they erode what made his unique and add noise between him and his customers. He now treats new projects more as a hobby than a business.

He is mostly right. But "building is commoditized" and "you can't make money" are different claims. The numbers show which one holds.

Supply doubled, demand didn't

The App Store is on pace for over a million new apps in 2026, while downloads grew about 2%.

Per Sensor Tower, new App Store apps bottomed at about 420,000 in 2022. They reached about 600,000 in 2025, up 30%. Then about 560,000 arrived in the first half of 2026 alone, nearly a full year's worth in six months (9to5Mac, TechSpot).

Demand barely moved. Downloads rose 3% in 2025 and about 2% so far in 2026. Each new app now gets roughly half the attention a new app got a year ago.

The composition matters too. Between March 2025 and February 2026, new iOS publishers grew 21% while recurring publishers shrank 7% (AI Weekly, citing Naavik). The flood is first-timers.

The power law was already brutal

Only about 1 in 30 subscription apps ever reaches $10,000 a month, and that was measured before supply doubled.

RevenueCat's data shows 17.2% of apps ever pass $1,000 a month. About 3.5% reach $10,000, roughly the line where one person can go full-time. One year after launch, the top 5% earn 200 times what the bottom quartile does (The Hustle).

Older developer surveys show the same shape: 2% of developers took 54% of all app revenue, while 88% shared 11% (GamesBeat).

Building costs zero; customers cost $1,200

Industry benchmarks put average B2B SaaS customer acquisition cost around $1,200, up roughly 60% in five years (Affninja, SaaSHero).

Run the math on a typical vibe-coded tool at $20 a month. At $1,200 per customer, you need 60 months of payment to break even. That is five years of retention for a product a competitor can clone this afternoon.

The economics flipped. Building used to be the expensive, defensible part. Now it rounds to zero, and everything the business needs to survive sits on the other side of the launch button.

Where the moat moved

The code is the commodity; what you wrap around it isn't. Each differentiator moves one of the numbers above:

• Distribution. An audience, a network, a channel you already own. It moves CAC from ~$1,200 toward near zero.
• Domain depth. Knowing one customer's problem better than they can explain it. Surveys found enterprise-focused developers 2–3x as likely to earn meaningful revenue per app (SmartCompany).
• Proprietary data. A feedback loop that improves with use and can't be cloned from your landing page. It moves retention.
• Trust. A track record and a person accountable for the result. In domains where mistakes are costly, "built in a day" counts against you.
• Embedded workflow. Integrations, data and habits inside a team's daily work. It makes the five-year breakeven reachable.
• Persistence. Still shipping and supporting in year two. The top 3.5% are mostly people who kept going for years.
• Building skill alone. Roughly zero now. It is what doubled the supply.

The test for any project is one question: what do I have that someone with Claude and a free weekend doesn't? If the honest answer is nothing, it is a hobby. If it is domain knowledge, data, an audience or trust, there is a business, and AI just made it cheaper to run.

What about ASI?

The thread closed on a different note: people at the frontier expect superintelligence within two years, and if that happens, neither your app nor your bank balance will matter.

That may be true, but it doesn't help anyone decide anything. "Nothing matters in two years" can be attached to every choice, and frontier timelines have a history of running early. If the next two years are roughly normal, the people who hold data, trust and a track record will be the ones still earning.

So plan for the normal case. For someone switching careers, the skill that pays now is less "I can build" and more "I understand one domain and its customers, and I can build fast for them."

A note on sources: App Store figures are Sensor Tower estimates as reported by the press. CAC figures come from industry benchmark aggregators, so read them as direction, not precision. The enterprise-vs-consumer split comes from older developer surveys.

SPR{K3 is a security research operation that pairs offensive vulnerability research with runtime behavioral defense. Defend is our runtime agent. To talk about a deployment, reach us at support@sprk3.com.