Blog

WordPress is entering its Drupal phase

By Phillip Clapham · October 1, 2026

WordPress isn’t going to die. Not this year, not overnight. What it’s going to do is shrink, steadily, for the next decade, until it looks like Drupal does today: still serviceable, still running a lot of sites, and almost nobody’s first choice for something new.

I’ve worked in WordPress for twenty years, so I’m not saying this to be edgy. I’m saying it because I’m planning client work around it.

What “the Drupal phase” means

Drupal was the serious CMS once. On W3Techs it ran about 2.3% of all websites in 2018. Today it’s about 0.6% (W3Techs). It didn’t die. The people who make new things stopped choosing it, and the market around it thinned out year after year.

That’s the phase I mean. Not a crash, a slow narrowing.

The numbers have started to move

For years WordPress looked unstoppable. On W3Techs it sat around 43% of all websites from 2022 into 2025, peaking at 43.6% in January 2025. As of October 1, 2026 it’s 40.2%, about three points down from a year ago (W3Techs history). W3Techs’ own sample shifted in that window too (its “no CMS” share rose), so read it as a direction, not an exact count. The HTTP Archive’s 2025 Web Almanac called it WordPress’s “first sustained slowdown” (Web Almanac 2025).

Commerce shows it more sharply. Store Leads counts 4.05 million live WooCommerce stores, down 8% year over year from a peak of 4.57 million at the end of 2023. Shopify is at 3.11 million and up 11% (Store Leads, September 2026).

It was already fragile

None of WordPress’s problems are new. The plugin stack is fragile. The security load falls on site owners. Twenty years of backward compatibility mean every new idea gets bolted onto something old. Page builders sit on top of all of it.

It kept growing anyway, because of inertia. Developers had learned its quirks and didn’t want to start over. The ecosystem fed itself: hosts, agencies, premium plugins and themes, each depending on the others. That house of cards was stable because everyone had a reason to keep it standing.

Then the house caught fire

Right when WordPress most needed its community pulling together, its leadership picked a fight with one of its biggest hosts. The timeline is public:

  • September 2024: Matt Mullenweg attacked WP Engine from the stage at WordCamp US, then banned it from WordPress.org.
  • October 2024: WP Engine sued. Ten days later WordPress.org took over the slug of the ACF plugin and forked it as “Secure Custom Fields”.
  • December 10, 2024: a federal judge granted WP Engine a preliminary injunction, finding it likely to succeed on its interference claim.
  • September 2026: the antitrust claims against Automattic and Mullenweg were revived. Trial is set for 2027 (Search Engine Journal).

The company itself went through it too:

  • October 2024: 159 people took a buyout offer rather than stay (TechCrunch).
  • January 2025: Automattic cut its sponsored contributions to WordPress core from about 3,500 hours a week to 45, and resumed them around late May (Automattic).
  • April 2025: 16% of the company was laid off (Automattic).
  • September 9, 2026: Automattic’s board voted to put Mullenweg on leave. Roughly 33 hours later he was back as CEO. No reason for the vote was given. The interim CEO and the chief legal officer were then fired, and the board members who voted against him have since left (TechCrunch, TechCrunch).

Here’s my read, and it’s only my read: you can’t innovate your way through a platform shift while your own leadership is fighting your own ecosystem. The facts above are the evidence. The conclusion is mine.

AI is the iron bar

Here’s the part I think a lot of people are underestimating. The moment AI-driven development became the default, WordPress’s long-term position broke.

It isn’t that WordPress has no AI story. It does: there’s an official MCP adapter and an Abilities API in core (WordPress AI team). The problem is underneath. WordPress is twenty-plus years of nested backward compatibility: hooks on hooks, plugins rewriting each other’s output, page-builder markup nobody wrote by hand. AI tools are at their best on clean, legible, file-based projects they can read end to end. WordPress will never be that, because it can’t break the millions of sites built the old way.

Frameworks built for how sites get made now don’t carry that weight. Astro is the clearest example: content as plain files, static output, nothing to patch, and official tooling for AI coding agents (Astro docs). In January 2026 Cloudflare acquired the Astro team. Astro stays open source and deploys anywhere, but it’s now backed by the company that already runs DNS, caching and hosting for a huge share of the web (Astro). Cloudflare has since released its own WordPress-style CMS on top of Astro, called EmDash.

Who gets squeezed

It goes beyond the CMS. WordPress’s market depth supports a whole economy: agencies, managed hosts, premium plugin and theme sellers, page-builder subscriptions. Their rates and margins exist because the market is that deep.

AI-first workflows change the maths. The agency that survives the next decade moves to AI-first workflows, on AI-first tooling, on AI-first platforms and hosting. It runs the whole thing with a fraction of the people and a fraction of the cost of a human-only WordPress shop. Along the way it stops paying for the managed host, the premium plugins, the premium theme and the page builder.

Each of those cancellations is small. Together they’re the house of cards coming down, slowly, and the last people on it are left in an ecosystem that looks like Drupal’s.

The one move I’d bet on

There’s one place AI-first platforms don’t yet cover well: commerce, where money, inventory, fraud and trust all meet.

Agents are starting to buy things. OpenAI and Stripe launched an agentic commerce protocol in 2025, Google announced AP2, and Shopify and Google co-developed UCP in January 2026. WooCommerce is in the game: it shipped an MCP integration and was a launch partner for Stripe’s agentic commerce tools. But it’s a participant, not the standard.

The version of this future where WordPress matters more, not less, has WooCommerce as the trust layer for agentic e-commerce: the governed, auditable core that agents transact through, with WordPress as an AI-friendly headless back end behind fast static front ends, and hosts as the managed layer for that governance.

I don’t see the will to build it. So I’m planning for the Drupal phase.

What I’d do with your site

Your site My advice
A brochure site Leave. A static site on Cloudflare costs close to nothing to host and has almost nothing to hack. You shouldn’t be paying for hosting at all.
A simple store Leave for Shopify.
A store with real custom needs Stay for now. Make it fast and stable, and plan.
A membership site Plan your exit. WordPress was never built to scale this.
Something broken right now Rescue it first, decide second.

If you’re a business owner reading this: nothing has to happen this week. But the switching cost only goes up from here.

Send me your site and I’ll tell you which row you’re in. [email protected]. If it’s broken right now, start with the WordPress rescue. If you’re ready to leave, here’s how a move works.