WordCamp Europe 2026: what Kraków says about the future of WordPress

I came back from Kraków with mixed feelings, and from a vantage point that obliges us to be honest: I co-created WordCamp Europe 2026 from behind the scenes, on the budget team, so I watched this event not only from the room but from backstage. As an event, WCEU was excellent: 2,458 ticket holders from 81 countries, the venue, the people, and the city itself are hard to grade as anything but top marks. It was worse with what hung in the air between the talks. If you hold that first impression from the room up against the hard numbers, the picture becomes coherent and uncomfortable: WordPress is not dying, but for the first time in decades it is genuinely shrinking, and a good part of the ecosystem already has one foot out the door.

#How WordCamp Europe 2026 went in Kraków

Contributor Day was full: nearly 800 people took part, good energy, concrete conversations. And yet two things cast a shadow. The first was the founder’s absence: Matt Mullenweg announced on his blog that he decided at the last minute not to fly to Kraków and did not appear in person, explaining it as a wish to stay close to loved ones during a difficult time; Mary Hubbard led the closing conversation with Matías Ventura and Rich Tabor. It is an understandable decision on a human level, but the leader’s absence from the community’s flagship event did not help the mood around the project’s leadership.

The second thing was subtler. In corridor conversations with product and agency owners, the same tone kept coming back: almost everyone is either looking beyond WordPress or already has one foot there. That is my impression from the room, not a survey, but when you lay it over the data it stops being just an impression.

#What was announced at WordCamp Europe 2026

The stage program did not avoid the questions I heard in the corridors. According to the official WordPress.org recap, the schedule held 49 talks and eight hands-on workshops, and close to a quarter of the ticket holders were at their first WordCamp Europe. Four points from the stage matter for anyone deciding where WordPress is heading:

• CERN’s web team used the keynote to announce that home.cern, the laboratory’s flagship site, now runs on WordPress after an automated migration.
• WordPress 7.0 ran through the whole conference. The release moved a native AI client, the Abilities API and a Connectors screen into core; what that means in practice is described further down.
• Alain Schlesser, a WP-CLI maintainer, pointed out that AI assistants and search already send real traffic to the open web, with more than a billion referral visits logged by the middle of 2025, and gave a checklist for making a site easy to find, read and cite.
• In the closing session Mary Hubbard, WordPress Executive Director, talked with Matías Ventura, the Gutenberg project lead, and Rich Tabor. Ventura demoed desktop mode and open canvas experiments that reimagine the admin, and pointed to Studio Code, an open source, agent-based coding tool built for WordPress. Kraków University of Technology announced a WordPress course starting in October.

On open source and AI, Hubbard argued that the values that made WordPress work should also shape AI: “We should be talking about it, and we should be much louder about it” (Mary Hubbard, quoted in the WordPress.org recap). For the question in the title, that is the clearest signal from the stage: the project’s leadership treats AI as the next layer of core, not as a plugin topic. The next WordCamp Europe takes place in Málaga, Spain, from 27 to 29 May 2027.

#Is WordPress market share below 42 percent in 2026

For years the industry repeated „WordPress is 43 percent of the web”. That sentence is no longer true. According to W3Techs, the share has fallen below 42 percent of all websites, and the monthly series shows a sixth consecutive decline: from 43.2 percent in December 2025 to roughly 41.9 percent at the end of May 2026. That is a drop of 1.3 points in half a year, roughly twice as much as across all of 2025. The HTTP Archive Web Almanac for 2025 describes the same phenomenon more cautiously: the first sustained slowdown after decades of expansion.

You can argue about the scale. A point or two of percent a year is still hundreds of millions of sites for the biggest player, and it is not a catastrophe. But the direction is unambiguous and the pace is accelerating, and that is what is new.

#Which platforms sites leave WordPress for

The most interesting thing is not that WordPress is losing, but that no one in particular is gaining from it. Rival builders grow minimally and steadily: over a year, Wix added a fraction of a point, Shopify less, Squarespace and Webflow less still. None of these systems shows the share instability that WordPress does. The share moves to the „no recognised CMS” category, which over the same period grew from around 28.6 to around 29.5 percent. Those are static sites, front ends built on frameworks like Next.js, which the October 2026 W3Techs reading puts at 3.7 percent of all sites, lightweight stacks like Astro, and increasingly sites assembled by hand or with AI that leave no trace of WordPress at all.

It is a dispersed exodus, not a switch to a single platform. That is why it is hard to fit into one headline and why it is easy to play down. But this very pattern, many fleeing in many directions, is more dangerous than losing share to one rival.

#How WordPress market share changed after WordCamp Europe

The figures above describe the situation at the time of Kraków. The monthly W3Techs series for all websites shows what happened next:

In the four and a half months after the conference WordPress lost another 2.2 points, more than the 1.3 points it lost in the six months before. The category without a recognised CMS gained 2.6 points over the same period, while Shopify and Wix barely moved. The pattern described above held: the share does not go to a rival platform, it goes to sites where the detector sees no CMS at all.

#How the WordPress decline affects plugin sellers and agencies

Here I reach a distinction that changes the whole picture and that came through more clearly in conversations than on stage. The decline hits the product side of the ecosystem hardest. Sales of commercial plugins and themes on marketplaces have been falling for years, and the number of new installs starting life on WordPress is far lower today than two or three years ago. If your business depends on bringing new users into a „free to paid” funnel for a low-value-add plugin, you have a real structural problem.

The service side looks different. Some agencies report their best years at the same time, and a few consciously go back to one hundred percent WordPress work, because at the service level the demand for well-built, maintained, and secure sites has not gone away. This is not a contradiction with the share data but its other side: the commodity market is shrinking, not the value market. For a site owner there is a simple conclusion in this. Stop asking whether WordPress is growing and start asking whether your provider delivers value beyond mere installation and whether they will cope when the project outgrows a single system.

#AI content management in WordPress in 2026

The second recurring thread was about content and AI. The consensus, including among people close to the project core, was surprisingly aligned: nobody yet has a good, canonical process for AI-assisted content management in WordPress. Some people consider their flow decent, but usually because they have not seen how an agent works on a local static site, where changes to content, structure, and metadata happen in bulk and instantly, without clicking through a dashboard.

There is also a healthy counterpoint worth taking with humility: not everyone needs AI to manage content. Sometimes it is faster to simply write and publish than to build a complicated pipeline for its own sake. But where the scale of content is large, the advantage of a modern stack with files in a repository and an agent running locally becomes hard to ignore. This is one of the areas where WordPress, despite work on the core and on AI interfaces, is currently lagging.

#What WordPress 7.0 changes for AI

WordPress 7.0 “Armstrong” was released on 20 May 2026, two weeks before Kraków, and it is the core project’s response to the gap described above. According to the developer notes on make.wordpress.org, four things landed in core:

• The AI client is provider-agnostic. A plugin starts every request with wp_ai_client_prompt(), and WordPress routes it to a model from a provider the administrator has configured. Core does not bundle any provider; Anthropic, Google and OpenAI ship as separate official plugins.
• The Connectors API adds a Connectors screen under Settings with three featured connectors. An API key is looked up first in an environment variable, then in a PHP constant, then in the database. Keys stored in the database are masked in the interface but not encrypted; encryption is being discussed in Trac ticket #64789.
• The wp_ai_client_prevent_prompt filter lets a developer block specific prompts. A blocked prompt makes no AI call and returns a WP_Error.
• The Abilities API arrived on the PHP side in WordPress 6.9, and 7.0 adds the JavaScript counterpart, so plugins can declare what they can do in a way other tools can discover.

This is infrastructure, not a finished editorial workflow. It gives plugins one shared way to talk to models and gives the site owner one place to manage keys. A canonical process for managing content with AI, the thing people in Kraków said was missing, is still left to plugins and teams. For a site owner the practical steps are smaller: keep provider keys in an environment variable rather than in the database, and check which plugins call the AI client before you connect a paid provider.

#How to read WordPress market share on W3Techs

W3Techs publishes two different numbers and they are easy to mix up. The first is the share of all websites, discussed above. The second is the share among websites whose CMS can be recognised, and it is much higher: in the reading of 15 October 2026 the service shows 58.5 percent among sites with a recognised CMS and 39.7 percent of all websites. When one person quotes “almost 60 percent of the CMS market” and another “40 percent of the web”, both can be right, because they measure different things.

The second issue is method. According to the W3Techs methodology, the sample covers more than 20 million sites considered relevant, excluding duplicate sites and redirected domains, and subdomains count as one site. All blogs on wordpress.com subdomains together count as one site in this statistic, and the same goes for wix.com. The share therefore describes independent websites, not the number of installations.

The third issue is arithmetic on the figures in this article. From December 2025 to May 2026 WordPress lost 1.3 points, while the category without a recognised CMS gained about 0.9 points. Roughly two thirds of the lost share went where the detector sees no platform, and the rest spread across builders and smaller systems. That is only an approximation, because sites do not move from one category to another one to one: some are created, some disappear. The direction still matches the dispersed exodus described above.

What WordCamp Europe in Kraków means in practice for agencies I cover separately in my article on WCEU 2026 and agencies.

#Should you stay on WordPress or migrate

The worst reaction to this data is panic and blind flight; the second worst is pretending nothing is happening. WordPress is not dying; it still powers hundreds of millions of sites, and for large sites, shops, and editorial operations it is often the most sensible choice. But the market has matured enough that carrying on out of habit has stopped being safe.

My conclusion after Kraków is pragmatic and boring, and therefore credible. Be on the side of value, not commodity. Build so that migration is an option, not a trauma: versioned code, clean data, no dependence on a single plugin. Work on the stack that fits the problem, not the habit, meaning WordPress where it wins on complexity and ecosystem, and a static stack with a modern front end where speed, maintenance cost, and AI content flow matter. This is exactly what I do day to day, and that is why I read WordPress’s declining share not as a threat but as the market reordering itself around what I considered right anyway.

If you are wondering which side of that line your project sits on, describe what it needs to achieve and I will assess together whether to stay with WordPress or prepare for migration.

#How to check whether your site is ready for migration

I came back from Kraków with the advice to build so that migration stays possible. You can check that on your own project before deciding whether to leave WordPress at all:

• Code: is the theme, with your own modifications, in a repository with change history, or does it exist only on the server.
• Data: does the content live in posts and fields that can be exported, or is it locked in the formats of a single plugin.
• Dependencies: which plugin is the only way to your data. If disabling it breaks how content is displayed, that is where the clean-up starts.
• Site type: a content-driven, simpler site is a candidate for a static stack with a modern front end, for example Astro or Next.js. A shop, a large portal and a newsroom usually stay on WordPress.
• Content scale: with a large number of texts, an AI workflow on files in a repository gives a real advantage. With a handful of pages it is faster to just write and publish.
• Provider: does your team work on more than one stack and will it cope when the project outgrows a single system.

If code, data and dependencies come out well, you stay on WordPress by choice, and you can plan a move for when it actually pays off.

This is analysis and my opinion, based on a first impression from WordCamp Europe 2026, which I helped create, and on public data and reports. The names WordPress and WordCamp are trademarks of their owners and I use them descriptively only.

Last updated: 15 October 2026.