Cloudflare experiment ports most of Next.js API 'in one week' with AI
The Register
February 25, 2026
AI-Generated Deep Dive Summary
A Cloudflare engineer has successfully implemented 94% of the Next.js API using AI, specifically Anthropic's Claude, in just one week with minimal costs. This experiment aimed to address challenges faced when deploying Next.js on platforms other than Vercel, which often require extensive adjustments due to Next.js' proprietary toolchain. The project highlights the limitations of Next.js deployment adapters and the difficulties of integrating platform-specific APIs during development.
The engineer, Steve Faulkner, utilized OpenCode, an open-source AI coding agent, to design the architecture and write nearly all the code for Vinext, a new open-source project. Vinext replaces Next.js' Turbopack with Vite, resulting in faster build times (up to 4.4x faster than Next.js 16) and smaller client bundle sizes (56% reduction). Despite its promise, Vinext is still in early development, lacks human code review, and currently supports only Cloudflare Workers as a deployment target.
This experiment underscores the potential of AI in accelerating software development, particularly for projects with extensive documentation and test suites. While the project is not production-ready, it demonstrates how AI can enhance efficiency and scalability in building modern web frameworks. For developers and tech enthusiasts, this innovation offers insights into leveraging AI to overcome deployment complexities and improve tooling across platforms.
The success of Vinext could have broader implications for the future of AI-driven development, showcasing its ability to streamline processes and reduce costs while addressing real-world challenges in software deployment. As Next.js continues to evolve, such experiments highlight the importance of flexibility and interoperability in modern frameworks.
Verticals
tech
Originally published on The Register on 2/25/2026