Ben Thompson's vibe coding experiment asks: is software doomed?

Key Takeaways

• Ben Thompson's hands-on vibe coding experiment produced a functional app he plans to use daily, demonstrating AI coding accessibility
• Apple Intelligence remains blocked in Europe due to Digital Markets Act disputes, with implications for competitive dynamics
• Memory chip policy and China's growing capabilities raise geopolitical stakes that Thompson argues could affect Taiwan Strait stability

Ben Thompson spent part of June building an app to organize his garage. Not by writing code. By describing what he wanted to an AI and letting it handle the rest. The Stratechery founder calls this his "vibe coding adventure," and he's using it to stress-test a question that keeps surfacing in his analysis work: are software companies doomed?

Thompson's weekly roundup, published Friday, captures a peculiar tension in tech analysis right now. The questions feel enormous. Will white-collar work exist in a decade? Could chip export policy trigger military conflict over Taiwan? These aren't rhetorical flourishes. They're the actual stakes Thompson sees in AI development. Yet his practical advice is to stop fretting and start building.

What is vibe coding and why does Thompson care?

Vibe coding describes a workflow where you tell an AI what you want your software to do, then iterate on the output rather than writing code yourself. GitHub's 2025 survey found 92% of developers now use AI coding tools in some capacity. Thompson's experiment pushed further. He wanted to see if a non-engineer could produce something genuinely useful.

The result: an app he plans to use regularly. "Technology, for all of its importance, is also fun, and I'm having a blast," Thompson wrote. The garage organizer sounds trivial. That's the point. If AI can handle mundane software tasks, the implications for professional developers become uncomfortable.

Thompson published a detailed breakdown with ten takeaways from the experience. His position isn't that developers are finished. It's that the nature of the job is shifting. The competitive advantage moves from writing code to knowing what to build.

Apple blocks Siri AI in Europe over DMA fight

Buried in Apple's WWDC announcements was confirmation that Apple Intelligence, the company's full AI assistant suite, won't ship in Europe. The reason: Apple's ongoing dispute with EU regulators over the Digital Markets Act.

Thompson and John Gruber discussed the situation on their Dithering podcast, calling it "maddening." But Thompson's analysis goes further than frustration. He argues Apple's own policies around its ecosystem may create exactly the competitive changes the EU wants to see, just not through the mechanisms regulators intended.

European iPhone users get a worse product. Apple avoids regulatory risk. Neither side gets what it actually wants. This standoff could persist for years.

Memory chips, China, and Taiwan Strait risk

Thompson's coverage this week also flagged a longer-term concern. The three major memory chip manufacturers, Samsung, SK Hynix, and Micron, may regret opening doors to Chinese competitors. China's domestic chip capabilities are growing faster than export controls anticipated.

Microsoft's decision to use Chinese AI models adds another layer. Thompson sees incentive alignment: Microsoft benefits from model diversity and cost reduction, even as Washington pushes for decoupling. These crosscurrents make chip policy genuinely unpredictable.

The Taiwan question isn't hypothetical in this context. TSMC's advanced fabs remain the bottleneck for AI hardware. Any military scenario involving Taiwan would reshape the entire semiconductor supply chain overnight.

Figma's CEO on why AI helps incumbents

Thompson's interview with Figma CEO Dylan Field offered a counterintuitive argument. Field believes AI gives established design tools a tailwind, not a headwind. Companies with existing user bases can integrate AI features faster than startups can build complete products from scratch.

This matters for anyone evaluating software investments. The disruption narrative assumes AI helps newcomers. Field's position, backed by Figma's user data, suggests incumbents with strong workflows may actually consolidate their advantages.

Frequently Asked Questions

What is vibe coding?

Vibe coding describes building software by describing what you want to an AI, then iterating on its output, rather than writing code directly. The term became popular in 2025 as AI coding tools improved.

Why is Apple Intelligence not available in Europe?

Apple is withholding its full AI features from EU markets due to disputes over the Digital Markets Act, which requires interoperability and data-sharing that Apple considers a security and competitive risk.

How much does Stratechery cost?

Stratechery subscriptions run $12/month or $120/year for full access to Thompson's daily updates and podcast network.

What did Ben Thompson build with vibe coding?

Thompson built a garage organization app using AI assistance. He published ten takeaways from the experience and says he plans to use the app regularly.

Why do memory chip makers matter for AI?

High-bandwidth memory is essential for training and running large AI models. Thompson argues that Samsung, SK Hynix, and Micron may face unexpected competition from Chinese manufacturers despite export controls.

Source: Stratechery by Ben Thompson