In a single quarter of 2026, Apple's App Store received more new app submissions than it processed in some full years of the early 2010s. The supply side of software has changed shape entirely, and the pace is still accelerating.
The bottleneck is no longer how to build a product. The bottleneck is how to tell which product is worth buying.
Apple's Review Queue Is Buckling
In Q1 2026, Apple processed roughly 235,800 new App Store submissions - an 84% year-over-year surge from Q1 2025. Apple's full-year 2025 number was 557,000 new apps (The Information, March 2026). Sensor Tower's monthly tracking puts the December 2025 growth rate at 56% and January 2026 at 54.8%, the highest figures in four years.
Apple's review pipeline was sized for around 200,000 submissions per week at peak. It is no longer hitting that target. Review times have stretched from a historical 24-48 hours to somewhere between 7 and 30 days. Apple has begun pulling entire vibe-coding platforms from the store: prompt-to-app builder Anything was removed on March 30, 2026 (Guideline 2.5.2); Replit and Vibecode have faced similar blocks.
Why The Pipe Is Filling Up
The cause is "vibe coding" - software built primarily by natural-language prompting of an AI rather than written line-by-line. The market was sized at $4.7 billion in 2025 and is projected to reach $12.3 billion by 2027, a 38% CAGR. The term itself was coined by Andrej Karpathy in February 2025 and named Collins Dictionary's Word of the Year that same year.
Underneath that label, the inputs are moving fast:
- 46% of all new code committed globally was AI-generated in Q1 2026 (GitHub).
- 84% of developers use or plan to use AI coding tools, up from 76% a year earlier (Stack Overflow Developer Survey 2025, n=49,009 across 177 countries).
- 92% of US developers use AI tools daily.
The Builders Have Changed
The most important shift is who is actually using these tools. The Second Talent Industry Report 2026 found that 63% of vibe-coding users are not professional developers - they are operators, designers, marketers, founders, and domain experts shipping production software for the first time. Y Combinator's Winter 2025 batch contained startups whose codebases were 95% AI-generated in roughly one in four cases.
Deloitte's 2025 forecast projects the population of "software creators" growing from around 30 million professional developers today to over 100 million citizen developers by 2028. Gartner expects 75% of new application development to be low-code or natural- language by 2026, up from 40% in 2021.
The Tools Are Compounding
The platforms enabling all of this are themselves growing at rates without precedent:
- Cursor (Anysphere): $2B annualized revenue by February 2026, $29.3B valuation, 7M monthly actives. The fastest SaaS company in history to reach $100M ARR.
- Lovable: $400M ARR in twelve months from launch, $6.6B valuation, over 25M projects, more than 200,000 created daily.
- Replit: $24M to $240M after the September 2024 launch of its AI Agent. $400M Series D at $9B in March 2026.
- GitHub Copilot: over 20M users, 4.7M paid subscribers, deployed at 90% of Fortune 100 companies.
Israel sits inside the same wave. Wix acquired Base44 - a one-founder AI app builder with no outside funding - for $80 million in 2025. Israeli startups raised $15.6 billion the same year, with roughly 70% of that capital flowing into AI and cybersecurity.
What This Means For Buyers
For anyone looking to buy software rather than build it, this matters in a specific way. The market is no longer scarce. Every category a buyer might want now has five, twenty, or two hundred indie products competing for it. Most are made by people the buyer has never heard of. Some are excellent. Some are abandoned within ninety days. Almost none come with the institutional context a buyer of, say, an enterprise SaaS would have taken for granted three years ago.
Supply solved itself. Trust didn't. The rest of this blog is about that gap and what fills it.
