When Cloudflare Rewrites Next.js: The Clash of Models in Open Source

In November 2024, Cloudflare announced that it had rewritten a significant portion of Next.js to run on its Workers infrastructure. A technical decision on the surface, but one that actually reveals a growing tension in the open source ecosystem: the gap between the promises of freedom and commercial reality. At the same time, generative AI is redrawing the boundaries of what had been a fragile balance between community contribution and monetization.

This convergence is hardly trivial. It signals a shift in how tech companies build, distribute, and control their tools. Understanding what's at stake here means anticipating the infrastructure on which we'll build applications over the coming years.

Cloudflare's rewrite of Next.js: a technical move with strategic implications

Next.js, developed by Vercel, has established itself as the go-to React framework for building high-performance web applications. Its massive adoption rests on a classic model: a free open source tool, an active community, and a premium hosting platform that monetizes the infrastructure optimized for this framework.

The problem? Next.js is designed to run optimally on Vercel. Advanced features, particularly server-side rendering and route management, rely on specific infrastructure paradigms. Cloudflare, whose Workers architecture is built on a different model (V8 isolates rather than containers), faced a dilemma: accept degraded performance or take control.

Cloudflare's team chose the latter. Rather than adapt their infrastructure to Next.js's constraints, they rewrote the incompatible parts to run natively on Workers. This move is significant: it means there are now two diverging implementations of the same framework, each optimized for its proprietary ecosystem.

From a technical standpoint, you could applaud the initiative. Cloudflare improves performance, reduces costs, and offers a viable alternative to developers. But from a strategic perspective, this fragmentation raises questions. Who ensures these two versions remain compatible? Who arbitrates future evolution? And most importantly, what becomes of the promise of interoperability that underlies open source? A situation that recalls the risks of over-dependence on a single ecosystem.

Commercial open source caught in its contradictions

The commercial open source model rests on a bet: offer software for free to drive adoption, then monetize through premium services, support, or proprietary infrastructure. This model has produced undeniable successes, from Red Hat to MongoDB to Elastic.

But it also creates structural tensions. Companies that publish open source code must navigate between two contradictory imperatives: nurture a community that expects transparency and sharing, while protecting their competitive advantage to justify their valuation.

Vercel finds itself in this uncomfortable position. Next.js is technically open source, but its development is driven by commercial logic that favors integration with the Vercel platform. External contributions exist, but major decisions remain centralized. This concentrated governance naturally creates frustration among players who want to use Next.js in different contexts.

Cloudflare, in rewriting Next.js, exposes this contradiction. If code is truly open, why not accept that it be adapted, forked, reinterpreted? But if every major player creates their own version, what's left of the promise of a shared standard? We're witnessing a classic phenomenon in software history: fragmentation follows mass adoption, and commercial interests ultimately prevail over community cohesion.

Open source licenses under pressure

Facing these tensions, several companies have hardened their licenses in recent years. MongoDB switched to the SSPL license to prevent cloud providers from offering their database as a hosted service without returning value. Elastic did the same with its proprietary license after Amazon launched a competing service based on Elasticsearch.

These license changes reveal a fundamental asymmetry: small businesses and individual developers benefit from open source code, but cloud giants can integrate it into their massive infrastructures and capture most of the economic value without contributing significantly to development. Open source creators find themselves in a position where they're indirectly financing their competitors' growth.

Generative AI as an accelerator of fragmentation in open source

The emergence of language models capable of generating code fundamentally changes the playing field. These tools don't just assist developers: they redefine the very notion of contribution and intellectual property in open source.

Take a concrete example. A developer uses GitHub Copilot, trained on billions of lines of open source code, to generate a feature. Is this code, inspired by thousands of existing projects, truly new? Should it be subject to the same licenses as the projects that trained the model? Do the maintainers of those projects have a right to recognition, or even compensation?

These questions aren't theoretical. Several lawsuits are underway in the United States to determine whether training models on open source code constitutes fair use or a license violation. The outcome of these disputes will largely determine the future of open source in the age of AI.

But beyond legal aspects, AI alters the dynamics of contribution. Historically, open source rested on an implicit social contract: I share my code, you improve it, the community benefits. AI short-circuits this cycle. It consumes existing code to generate new variations, without necessarily enriching source projects. We're shifting from a collaborative contribution model to a one-way extraction model.

Productivity as a new barrier to entry

AI also multiplies the capacity of large enterprises to produce code quickly. An engineer assisted by generative tools can maintain multiple open source projects in parallel, create alternative implementations, adapt existing frameworks. This productivity gain, concentrated among players with access to the best models and infrastructure, amplifies the asymmetry between tech giants and individual contributors.

Cloudflare, in rewriting Next.js, likely benefits from these tools. It becomes technically easier to create an alternative version of a complex framework when you can leverage models capable of translating code, optimizing it, generating tests. This technical ease could encourage other players to follow suit, multiplying forks and diverging implementations.

Toward a new balance between openness and sovereignty

Facing these upheavals, several scenarios are emerging. The first, pessimistic, would see open source fragment into incompatible silos, each controlled by a commercial player seeking to lock in their ecosystem. Developers would find themselves in a paradoxical situation: more apparent choices, but less real interoperability.

The second scenario, more optimistic, would rest on the emergence of new governance models. Independent foundations, like the Linux Foundation or Apache Software Foundation, could take control of strategic projects to guarantee their neutrality. These structures would allow separating software development from commercialization, creating a space where contributions are evaluated on technical merit rather than alignment with a commercial strategy.

A third scenario would involve more active regulation. The European Union, with the Digital Markets Act, has already laid the groundwork for an approach where dominant platforms must guarantee interoperability. We could imagine this logic extending to frameworks and development infrastructure, imposing technical standards to prevent lock-in situations.

What this means for tech teams

For organizations building applications on these technologies, the implications are immediate. Choosing a framework can't be reduced to technical evaluation anymore. You must now incorporate a strategic dimension: what's the project's governance? Who controls its roadmap? What's the likelihood of future fragmentation?

Teams must also anticipate AI's impact on their own code production. Generative assistants become standard tools, but they introduce new responsibilities: verify license compliance, audit generated code, maintain deep understanding of systems rather than blindly relying on automatic suggestions. An approach similar to the supervision necessary in any AI system in production.

The Cloudflare-Next.js affair is merely a symptom of a deeper transformation. Open source, as conceived in the 1990s and 2000s, rested on assumptions that no longer hold: individual contributors motivated by altruism, companies that monetize through support rather than infrastructure, clear separation between software and service. AI and cloud concentration challenge each of these pillars.

The question is no longer whether open source will change, but how it will reinvent itself. Decisions made today by players like Cloudflare, Vercel, and AI model creators are shaping the contours of the tech ecosystem for the next decade. Ignoring these dynamics risks building on foundations that could fragment beneath our feet.