The lifeline is called AI
WordPress 7.0 has arrived, and with it what the project leadership is selling as the biggest step since the introduction of Gutenberg: a native AI layer that brings Gemini, Claude and ChatGPT directly into the backend. Generate content, suggest layouts, create images, fill in fields, assemble entire pages from a prompt – all of it now runs not through a flaky third-party plugin but as an integral part of the system.
On paper, this is the obvious answer to an existential question. If modern frameworks have outpaced WordPress in performance, security and developer experience for years, then the CMS has to score somewhere. And AI is the card everyone is playing right now. The logic: anyone pulling content into WordPress at the push of a button from an AI model no longer needs a page builder, no longer needs ten plugins, no longer needs an agency for every landing page.
The logic isn't wrong. It's just too short-sighted.
What the AI integration actually clears away
The first, often overlooked effect of WordPress 7.0 doesn't hit the competition – it hits its own ecosystem. Over 60,000 plugins and countless theme shops exist because WordPress can do little at its core and bolts everything on through extensions. A large share of these plugins solves problems that a capable AI layer makes trivial: building forms, writing SEO copy, generating meta descriptions, captioning images, translating content, producing layout variants.
When the model in the backend takes over these tasks directly, the economic basis for a large part of the plugin and theme market collapses. Why buy an SEO plugin with annual licence fees when the integrated model delivers the same output? Why a premium theme when the AI assembles a layout on demand?
That's the real explosive force of 7.0: it points inward. In the medium term, the AI integration means the end of the plugin and theme ecosystem as we know it – and that ecosystem was, for over twenty years, the only real moat WordPress had. It wasn't the technology that kept WordPress alive, but the network of developers, agencies and marketplaces that make a living from it. Drain that market, and you saw at your own foundation.
You can't lock the AI in a cage
The second, strategically decisive fallacy lies in the architecture of the integration. WordPress is trying to embed AI as a controlled feature within its own system – as a tool that sits obediently in the backend and waits for commands. Generate content, yes. But inside the WordPress world, by WordPress rules, rendered through the WordPress pipeline.
This misjudges what is actually happening. AI tools aren't a feature you build into an existing system. They are the layer from which software emerges in the first place. Developers today build complete websites in a fraction of the time by instructing a model – not in WordPress, but in a modern framework that the AI handles far more cleanly than nested PHP with thirty layers of hooks.
Bringing AI into WordPress in order to tame it is like trying to catch a flood with a bucket. The models have long since taken control of the process by which software comes into being. The question is no longer "How do I integrate AI into my CMS?" but "Do I even still need the CMS if the AI can generate the frontend directly?" For more and more projects, the answer is no.
The real problem stays untouched
Even if you set both points aside – the market cannibalisation and the false assumption of control – the hardest objection remains: AI in the backend changes nothing about the technological substance of WordPress.
An AI that generates content doesn't change the fact that this content ends up rendered through the same server-side PHP pipeline that has remained essentially unchanged at its core since the 2000s. It doesn't change the single MySQL database abused for everything. It doesn't change the attack surface, the caching crutches, the plugin compatibility conflicts, the load times of 800 milliseconds to several seconds.
A text generated by AI inside a slow, insecure, hard-to-maintain system is still a text inside a slow, insecure, hard-to-maintain system. The AI beautifies the surface of content creation while the foundation stays the same. It's like fitting a combustion-engine car with a modern infotainment system and selling it as an electric vehicle. The display is impressive. The drivetrain is not.
Modern architectures solve the problem at the root: static site generation and edge rendering deliver pages in under 100 milliseconds, with no server collapsing under load. Headless setups have no public admin interface as an entry point. There, AI isn't a feature bolted on after the fact but part of the development process from the start – cleanly typed, versionable, traceable.
Why the reflex is understandable nonetheless
The WordPress maintainers' decision shouldn't be dismissed as foolish. It's a rational response to real pressure. Market share has stagnated for years, new projects increasingly go to Next.js, Astro and the like, and AI is the only topic that currently generates attention and a sense of momentum. From the perspective of a mature product losing relevance, the AI card is almost the only one to play.
The problem isn't the will to renew. The problem is that it answers an architecture problem with a feature. You can repaint a house with cracks in its foundation as beautifully as you like – the cracks remain. And the AI integration is exactly that: a fresh coat of paint on a foundation that no longer holds.
What this means for companies in concrete terms
For companies sitting on WordPress today and hoping 7.0 will be their modernisation, the sober assessment matters. The AI features will make editorial work easier – that's real. But they are no reason to keep postponing an architecture decision that is already due. On the contrary: anyone investing now in the WordPress AI world, building workflows on it and training editors on it, is increasing the switching costs for later.
Our recommendation stays the same as before, only with heightened urgency: migrate step by step rather than retrofit. First lift the public-facing, business-critical pages onto a modern architecture, then switch the content workflow, finally shut down the backend. This is precisely where AI pays off twice – not in the old system, but in the migration: AI-assisted tools today reduce the cost of a switch to a fraction of what it would have meant three years ago.
In projects we regularly see that the fear of migration is greater than the actual effort. What used to be a months-long major project is, with today's tools, often a matter of weeks – including content transfer, redirect mapping and SEO preservation.
Conclusion
WordPress 7.0 is not a comeback, it's a symptom. The AI integration solves two problems WordPress never had – writing content faster and building layouts faster – and ignores the one it has had for twenty years: an architecture from a different web era. At the same time, it dries up the plugin and theme market that kept the system alive in the first place.
You can't lock AI tools in a cage and hope they'll preserve the old world in there. They have already taken control of how software comes into being – and they prefer to emerge where the architecture fits them. WordPress isn't one of those places. 7.0 won't stop the decline. It will just make it look a little shinier.