What Forrester found
Cloudflare scores the maximum across developer experience, performance and latency optimisation, security, and AI application development. Its strategy scores are equally strong: perfect marks for vision, innovation, and roadmap. The report describes Workers as delivering an exceptional developer experience with zero cold starts, automatic workload placement for optimal latency, and a 100% uptime SLA.
The framing Forrester uses is worth pausing on: edge development platforms are “the happy path beyond the CDN.” That language captures something I wrote about in an earlier piece on Cloudflare’s edge-native direction. The shift from thinking about edge as content delivery to edge as a compute substrate is now the formal analyst position, not just a practitioner observation.
Before getting to the AI story, it is worth being clear about the base case. Cloudflare’s perfect scores on developer experience, performance, and security make it a strong infrastructure choice for any modern website, with or without AI in scope. Global distribution as a default, zero cold starts, a clean deployment model, and a security layer that does not require specialist configuration. These are reasons to build on Cloudflare for a standard web project. And when AI does come into the picture, the platform is already wired for it. Everything you need to build intelligent experiences comes built in: model routing, edge inference, agent workflows. Solid infrastructure with AI included, not bolted on afterwards.
The AI angle is not a footnote
What stands out in this Wave is how central the AI story has become. Cloudflare’s maximum score on AI application development is not a reward for having an AI feature on a roadmap slide. It reflects a platform that is genuinely repositioning edge infrastructure as the right place to run AI workloads.
The pieces they have assembled tell a coherent story:
AI Gateway: when your application talks to an AI model, this layer sits in between. It handles routing across model providers, keeps costs in check through caching, and gives you a clear view of every AI call your product makes. One control point for everything AI-related.
Replicate acquisition: Cloudflare now brings a wide catalogue of AI models directly into its platform. Choosing a model and running it happen in the same place. No separate integrations, no extra infrastructure to manage.
GPU expansion: generating an AI response takes compute. By running that on servers close to your users rather than in a distant data centre, response times drop noticeably. The logic is the same as for any other web performance improvement: proximity matters.
Durable execution primitives: AI agents do not just answer a single question. They work through steps, wait for input, and pick up where they left off. Cloudflare has built the underlying infrastructure to support that kind of multi-step behaviour reliably and at scale.
Forrester describes this as an “emerging agent-first development ecosystem,” which is accurate. It is still emerging. But the infrastructure underneath it is not speculative. We are already using parts of it, and the building blocks are in production.
The practical read: if you are building AI-assisted experiences that need to be fast, globally distributed, and cost-efficient at scale, the edge is the natural deployment target. Cloudflare is, right now, the most complete platform for doing that.
What we have seen in practice
Working with Cloudflare Workers in real projects, the experience matches the report’s findings. The deployment model makes global distribution a default rather than something you configure after the fact. The developer experience is clean. The security primitives are solid without requiring deep specialist knowledge to apply correctly.
The AI Gateway in particular is something we have looked at closely for clients running multiple model providers. The ability to route, cache, and get visibility into LLM calls from a single control point is genuinely useful, especially as AI costs at scale become a real budget line.
When I compared Cloudflare and Vercel directly in a previous insight, the distinction that matters most is how much freedom you retain. Vercel is deeply tied to Next.js. It is well-executed if that is your stack, but it is an increasingly closed system: opinionated about how you build, how you deploy, and what things cost as you scale.
Cloudflare takes the opposite approach. The platform is open, supports a wide range of runtimes and frameworks, and does not steer you towards a specific stack. You get more control over your architecture and more predictable costs as traffic grows. If vendor lock-in is a concern, or if Next.js is not the answer to every question in your project, that distinction matters more than it might first appear.
For companies building digital products, the practical shift is this: global distribution, low-latency responses, and AI-ready compute no longer require specialist infrastructure work. They come as defaults. The question is whether you are building on a platform that treats that as a first principle.
Interested in how this could work for your organisation?
If you are evaluating edge platforms for an upcoming project, working through where AI inference fits in your architecture, or trying to make sense of how Cloudflare relates to what you are already running, I am happy to dig into the specifics with you. The Forrester report is a useful reference point. The day-to-day experience of building on these platforms tells you a great deal more.