AI Vibe Coding Tools See Traffic Drop After Summer Surge (hyper.ai)

AI-powered “vibe coding” tools — platforms that let users build websites and apps from plain-text prompts — saw a sharp interest spike in spring/early summer and then a meaningful traffic decline, Barclays’ analysis of Google Trends and SimilarWeb data shows. Top names affected include Lovable, Vercel’s v0, Bolt.new, and Replit. Analysts warn that much of the early ARR growth may have been driven by transient, month-to-month subscribers rather than sticky customers; sector churn is high, confirmed by Bolt.new’s CEO. Compounding the problem are “inference whales” — a small set of heavy users that drive outsized AI compute costs, forcing startups to raise prices and impose usage limits to protect margins, which in turn can depress growth and retention. The drop matters because it exposes core economic and technical challenges for AI-native dev tools: production-readiness still demands the “final 5%” polish that non‑technical users struggle to achieve, and model inference costs plus supply/demand pricing dynamics strain unit economics. The space remains valuable for rapid prototyping among AI-native early adopters, and incumbents are responding (Wix acquired Base44; GoDaddy is exploring). Barclays and others see vibe coding as evolving rather than dead, but widespread adoption will require better retention, cost controls (throttling, smarter pricing, model optimization), and stronger tooling to bridge prototyping to production — a transition that could take until around 2026.