App Store New Apps Surge: AI Coding Tools Behind 60% Spike

App Store New Apps Surge: AI Coding Tools Behind 60% Spike

New iOS app submissions broke out of a three-year freeze in 2025, and AI coding tools are the most credible explanation. The number circulating in recent coverage, an 84% annual surge, has no primary source behind it. The verified figures are still striking, and understanding the difference matters for anyone trying to read where Apple's platform is actually headed.

New iOS apps grew 24% across all of 2025 and spiked 60% year-over-year in December alone, ending a stretch of essentially flat app creation that had held since 2022, per analysis drawing on Sensor Tower, Wells Fargo Securities, and a16z data reported by The Decoder in January 2026. The 84% figure has no disclosed methodology and no primary source to validate it; treat it as unverified until that changes.

The practical stakes go beyond one number. If AI coding tools are genuinely lowering the barrier to shipping iOS software, the platform may be entering a phase it hasn't seen since 2008: a supply flood. What that means for App Store review, search rankings, spam volume, and app quality depends entirely on whether that new supply is any good.

One clarification worth establishing upfront: throughout this piece, "new apps" means new listings in the App Store. Not downloads, not active installs, not revenue. Supply and demand are different data points, and conflating them is exactly how misleading App Store narratives get built.

The App Store 84% surge claim doesn't hold up, but what's underneath it does

The best-documented figure here is a 60% year-over-year jump in new iOS app submissions in December 2025, with 24% growth across the full year drawn from Sensor Tower and Wells Fargo Securities, as summarized by The Decoder in January 2026. That's the verified baseline. No primary source reviewed for this article supports the 84% annual figure spreading through subsequent coverage.

The three-year plateau is what makes 24% meaningful. App creation on iOS had been effectively static since 2022. A mid-single-digit uptick in that context would be noise; a 24% annual increase followed by a 60% monthly spike is a structural break, not a blip.

Apple does not publish submission volumes, rejection rates, or category breakdowns. That opacity matters: the surge is visible in aggregate listing counts from third-party intelligence firms, not in any Apple-issued transparency report. How many new apps cleared review on the first attempt, which categories are growing fastest, what percentage are still live six months post-launch, none of that is available.

When any outlet reports on App Store growth, three questions separate signal from noise:

• Does the figure come from Apple directly or from market intelligence aggregation?

• Does it measure submissions, live listings, or downloads?

• Is there demand-side data, such as ratings, downloads, or retention, accompanying the supply claim?

Those three questions distinguish a growth story from a volume story. The current evidence answers only the first layer.

Why AI coding tools may be driving App Store growth

The leading explanation for the submission surge is the rise of "agentic coding," AI tools that generate functional software from plain-language descriptions, shifting the developer's role closer to product manager than programmer. For some simpler apps, these tools appear to be reducing the skill threshold from writing code from scratch to specifying product behavior. The infrastructure timeline makes this more than a convenient narrative.

Anthropic open-sourced the Model Context Protocol in November 2024, creating a shared standard for connecting AI systems to external tools and data. OpenAI, Google, and Microsoft all announced MCP support or integrations between March and May 2025. Anthropic launched its Claude Connectors Directory in July 2025. OpenAI pushed its app marketplace at DevDay in December 2025, the same month iOS submissions hit their sharpest monthly spike, per the timeline assembled by AdamGTM in March 2026. The timing lines up unusually well.

A16z drew an explicit parallel to 2008, when the original App Store launched with 500 apps and hit one million downloads within a weekend, as noted by The Decoder. The analogy has appeal. It's also where the parallel should stop. Demand conditions in 2008 were categorically different from an App Store already carrying millions of existing listings competing for attention.

The causation caveat belongs here, stated once, clearly: The Decoder's reporting is explicit that the data does not prove AI tools caused the surge. There are no developer surveys, no submission-level metadata confirming AI tool usage, no category-level breakdown. AI coding is the most structurally coherent explanation given the timing. It is not an established fact, and this article treats it as the working hypothesis, not the conclusion.

Alternative explanations the data can't rule out

Before leaning too hard on that hypothesis, it's worth naming what else could be driving the numbers. A post-2024 macro recovery in startup activity may have released a backlog of projects that had stalled during the 2022-2023 funding drought. Changes in how third-party tracking firms count listings could inflate apparent growth in aggregate figures without reflecting real submission volume. A category-specific spike, say in games or productivity, could be pulling the overall number upward in ways the aggregate hides. None of these alternatives can be confirmed or dismissed with the data currently available.

The AI coding hypothesis still fits better than any of them. It has a plausible mechanism, a documented infrastructure buildout, and a timing match that other explanations don't share. But the honest position is that it remains a hypothesis, not a verdict.

What low-barrier supply surges actually produce: a parallel worth examining

The App Store isn't the only software marketplace absorbing an AI-driven supply increase, and the others offer a useful preview of what happens when creation costs collapse faster than quality signals can adapt.

Claude's connector directory and ChatGPT's app listings each held roughly 150 products as of early March 2026 and had grown approximately 30% in the three weeks prior, per AdamGTM. The two platforms are already diverging structurally: Claude's 162 connectors are 79% pure B2B, anchored by enterprise GTM tools including HubSpot, ZoomInfo, Apollo, and Clay, with seven added in those same three weeks. ChatGPT's 142 apps skew 70% consumer. Same underlying infrastructure, different developer bases, different product priorities.

That divergence echoes an older failure. OpenAI launched ChatGPT Plugins in March 2023 with 11 integrations; they were dead by early 2024. The reasons were structural: most products were thin prompt wrappers, discoverability was weak, and the revenue-sharing model that had been implied never materialized, according to AdamGTM's retrospective. Volume came first. Utility didn't follow.

MCP gives the current wave a more durable technical foundation. It's a shared standard, not a proprietary plugin layer, and Sam Altman has framed the push as a path to reaching hundreds of millions of users through ChatGPT's roughly 800 million weekly active users, per AdamGTM. Whether that distribution access translates into products that stick is the same question the GPT Store failed to answer.

The direct parallel to the App Store is imperfect. Apple's platform has 17 years of infrastructure, review processes, and developer economics that AI assistant marketplaces lack entirely. But the pattern holds: when barriers to creation drop sharply, supply accelerates, quality control lags, and discoverability degrades before it improves. That arc may be what the App Store is entering now.

What a supply flood means for AI-generated apps on the App Store

More apps is not inherently bad news for Apple's platform. The question is what kind of apps. The available evidence on AI-assisted code quality suggests a meaningful fraction of new submissions may introduce security problems their authors don't know about.

Researchers generated 1,689 programs across 89 scenarios using GitHub Copilot and found roughly 40% contained security vulnerabilities, according to a study published in Communications of the ACM in February 2025. The mechanism is structural: AI coding tools are trained on public repositories, which contain insecure patterns, and reproduce them at scale.

The confidence gap may be the sharper risk. In a separate 47-participant study, developers who used AI assistance wrote less secure code than those without it on four of five tasks, and were more likely to rate their own output as secure (Kumar et al., October 2023). Developers who don't know what they don't know are less likely to flag problems during review, which is precisely where the damage gets through.

These studies cover general AI coding tools, not iOS-specific workflows or current agentic systems. They are risk indicators, not evidence that App Store submissions are compromised. The directional case is sound regardless: a wave of first-time developers using AI to build iOS apps likely includes a wave of submissions from people who cannot independently audit what those tools produced.

Apple's review process was designed in an era when shipping an app required enough technical investment to function as a natural filter. If AI removes that friction at scale, the review system faces a different kind of pressure, not just higher volume, but higher variance in quality, security, and intent. Spam detection, template-matching, and ranking algorithms calibrated to a slower-moving supply curve may need to be rebuilt for new iOS apps built with AI in mind. Nothing in the publicly available evidence suggests that recalibration has happened yet.

What meaningful demand evidence would actually look like:

• Ratings velocity on apps submitted in 2025 compared to prior cohorts

• Download-to-rating conversion rates for new developer accounts

• Featured placement rates for first-time developers

• Whether App Store search results for common queries become noticeably noisier over the next two quarters

None of that data is currently public. Those are the metrics that will distinguish a platform entering a new creative era from one absorbing a spam surge.

The number that matters, and the question Apple hasn't answered

The App Store submission surge is real, even if the 84% headline figure isn't. New iOS apps grew 24% across 2025 and 60% in December alone, a genuine break from three years of flat growth, per The Decoder's reporting on Sensor Tower and a16z data. AI coding tools remain a hypothesis, not a proven cause. But the infrastructure timeline, MCP adoption, Claude Connectors, OpenAI DevDay, gives that hypothesis structural weight that the alternatives don't easily match.

The security and quality risk is the part that doesn't get resolved by time alone. GitHub Copilot-generated code in one widely cited study showed roughly a 40% vulnerability rate in controlled research (ACM, February 2025), and developers using these tools consistently overestimated the security of their output (Kumar et al., 2023). Whether those risks show up in App Store submissions at scale is unknown. Whether Apple's review infrastructure is ready for them is also unknown.

The "App Store in 2008" framing is doing a lot of rhetorical work in current coverage. In 2008, the constraint was supply: barely enough apps existed to stock the shelves, and demand was essentially unlimited. Today the constraint is discovery and trust in a store with millions of listings already competing for attention. Adding more apps quickly doesn't solve that problem; it compounds it.

The question Apple hasn't answered publicly is whether it sees a supply surge as an opportunity or a stress test.