I built my first websites with Astro in early 2026. At the time, I couldn’t find anyone else doing it. I’d search for it and find a couple of YouTube videos. That was about it.
Then, sometime in October 2026, it went everywhere. It’s all over my LinkedIn and my X feed. Every day there’s a new post that starts with “We built our new website, and it’s not on WordPress,” or “Here’s how we moved 40 of our clients’ sites off WordPress.” And for the first time in a long time, the articles are about WordPress losing ground.
It looks like a vertical line. Nothing, then everywhere.
That’s what I see in my own feeds, though, so I checked it against the numbers. They mostly back it up, with a twist: the people who build websites have moved fast. The people who own websites haven’t started searching for it yet.
So people are going to start asking the obvious question: Astro or WordPress? I don’t think that’s the right question. The real news is that WordPress isn’t the automatic answer anymore.
What Astro Is
Astro is a free, open-source framework for building content websites: the kind most businesses have, where people come to read, look and get in touch. Most pages are built ahead of time and handed to visitors as finished files, so there’s no database and no plugins working behind the scenes every time someone loads a page. That’s a big part of why Astro sites tend to be fast, and why there’s so little on them to attack. In October 2026, 72% of Astro sites passed Google’s Core Web Vitals on phones, Google’s measures of how fast and stable a site feels to real visitors. For WordPress sites, it was 49%.
In January 2026, Cloudflare, one of the companies that runs a big piece of the internet’s infrastructure, acquired the company behind Astro. Astro stays open source, and you can still host it anywhere. Cloudflare also said Astro is already the backbone of platforms like Webflow and Wix.
Let’s keep it in proportion, though. Astro runs on about 0.3% of websites. WordPress still runs on about 38.5% (both as of November 20, 2026). This is early. But early is exactly when it’s worth paying attention.
What the Numbers Say
My feeds aren’t data, so here’s the data.
HTTP Archive, which crawls millions of websites every month, found about 21,500 sites running Astro in January 2026. By October 2026, it found more than 55,000. That’s more than two and a half times as many in nine months, and the climb got steeper as the year went on.
The people building websites moved the same way. Developers download Astro from npm, the registry most web code comes from, and its monthly downloads went from 4.6 million in January 2026 to 34.8 million in October, by npm’s public counts. Downloads include automated installs, so that’s a gauge of activity, not a count of sites. But it grew more than twice as fast as Next.js, the most popular framework of its kind.
Search tells a different story. On Google Trends, searches for “astro framework” jumped the week Cloudflare bought Astro, peaked in late May 2026 and have fallen back since. Searches for “Astro vs. WordPress” and “WordPress to Astro” are still too small for Google to chart. One of the fastest-rising related searches is Claude Code.
Put those together and you get the real picture. The builders have moved, and a lot of them are building with AI. The business owners haven’t started asking yet. That’s what early looks like, and it lines up with my own timeline: quiet in early 2026, then everywhere by October.
What’s Actually Changing
Here’s 2026 so far, with my own moments in it.
For years, WordPress’s share of the web only went one direction. Not anymore. It ran 43.0% of all websites at the start of 2026 and 38.5% on November 20, 2026.
You’ll see different numbers for WordPress depending on who’s counting. W3Techs, the source most people quote, puts it at 38.5% of all websites, or 57.4% of sites that use a content management system. HTTP Archive’s monthly crawl, which counts a different set of sites, had it at 30.4% in October 2026. The 2025 Web Almanac put it at 64.3% of mobile sites that use a content management system. They don’t agree on the size, because they count different things. W3Techs and HTTP Archive both show it shrinking through 2026.
Most of that didn’t go to Astro. It went to hosted site builders and to sites built without any recognizable platform at all, which W3Techs now counts at 33% of the web. But that’s the shift I’m talking about: people are building websites in ways that don’t start with WordPress.
It isn’t only small sites, either. Owner.com, which builds websites and online ordering for restaurants, rebuilt its restaurant sites on Astro. Their menu pages went from loading in 5.4 seconds to 0.7 seconds on real phones, and they report revenue per restaurant went up more than 4%.
I think the reason it’s happening so fast is AI.
Websites Used to Be Built This Way
Making a website now feels a lot like making one 25 or 30 years ago.
Back then, if you wanted a website, you hired programmers. Every change went through them and waited in their queue, and it was expensive. But a fully custom site had no real limits. You were limited only by what you could think of and what your developers were able to build.
WordPress brought the price down for everybody. Plugins made sites easy to extend, and that was great. It was also limiting. You built around the plugins that existed, unless you paid for a custom plugin, which most people never did. And every plugin brought bloat and security risk along with it. WordPress has always been a double-edged sword. (I wrote about that side of it in In 15 Years, Almost None of My Clients Ever Edited Their Own Website.)
Building with Astro is a lot more like the old days: custom code, built for one business. The difference is that AI keeps the cost of building down, the same way WordPress did.
That last line in the diagram matters. AI doesn’t replace knowing what you’re doing. It’s what lets someone who knows what they’re doing build a lot more. (That’s the difference between this and vibe coding.)
The Moment It Clicked for Me
For me, it started with a restaurant website.
I’d been building it in WordPress for months, starting in the fall of 2025, to launch before the restaurant opened for the season in May 2026. I was using a page builder, so I was building the entire design system myself: every component, the back end, how the team would edit it, and making sure it would be easy for them to use. Then all the things every site needs: the search settings, redirects from the old site’s pages, the descriptions and structured data Google reads, accessibility. You name it.
In the middle of all that, a major new version of the page builder came out. Suddenly I had to rethink everything I’d already built and decide what to redo for the new version, with launch right around the corner. I was only a few weeks out and not as close as I needed to be. I would have had to rush it and leave things unfinished.
So I moved the project into Claude Code, the AI tool I use to write code.
I felt safe doing it because I wasn’t starting from zero. This wasn’t a day-one prompt like “I need a website, here’s everything you need to know.” I had months of planning and building behind it. The project was probably 75% of the way there, and that gave Claude an enormous amount of context.
Once I was working in it, I could see what was actually possible. It worked, and it added a level of polish and detail I wouldn’t have had the time or bandwidth to reach on my own.
Then I thought about the companies building operating systems and platforms that serve millions of people. Anthropic says Claude now writes more than 80% of the code it merges, and Google says AI generates 75% of its new code. And I looked at what I was making: a restaurant website. Even though it’s code, it’s mostly visual. For an AI model, that’s fundamental work. That’s when I felt safe.
That site and the others I’ve built on Astro since all scored 100 for accessibility and 100 for search basics on Google’s Lighthouse test when I ran it on November 20, 2026, and between 81 and 100 for speed on mobile.
WordPress Is About to Get Better, Too
This isn’t a eulogy for WordPress. If WordPress fits your business, I think it’s about to get a lot better.
AI speeds up WordPress work too. I already manage WordPress sites by connecting AI tools directly to them. I’ve patched abandoned themes and plugins myself instead of waiting for their developers to do it.
That help is needed. WordPress put out two security releases in three weeks (October 28 and November 18, 2026), and 91% of the new WordPress security holes found in 2025 were in plugins. Someone has to keep up with that, whatever the site runs on.
The Next WordPress May Not Look Like WordPress
Here’s a sign of how fast this is moving. Cloudflare is building EmDash, an open-source system for managing website content, built on Astro, and Cloudflare calls it “the spiritual successor to WordPress”. The latest version, released in November 2026, can bring in the content from an existing WordPress site.
I haven’t tried it yet, so I’m not recommending it. But it tells you where the big players think this is going. Whatever replaces WordPress for a lot of businesses might not look like WordPress at all. For plenty of sites, it might not be a system like that at all.
The Two Objections I Hear Most
“But I can’t edit it myself.” After about 100 client websites, I can tell you most owners never do. They run their businesses, and they call me when something needs to change, because it’s more efficient and it lets them focus on their customers. Most small businesses aren’t hiring their own web developers, either.
If you really do want to edit it yourself, you still can. An editor can be built into an Astro site, and in a lot of cases it’s better than WordPress’s, because it’s built for your site. There’s none of the clutter of the WordPress dashboard, and nothing I have to lock down to keep you from breaking things by accident. It’s simple enough that anyone in your office could use it. That’s if you need one at all.
“I’ll be stuck with one developer.” In a way, you already were. WordPress’s real advantage here is that it’s so widely used that handing a site to another developer is easier. But every site has details the next developer doesn’t know. More often than not, when a client does move on (which doesn’t happen much), the next designer just rebuilds the site in WordPress, because that’s easier than learning what was already built. I’ve done the same thing when I’ve taken over other people’s sites. Sometimes a fresh start is easier than untangling someone else’s work.
Here’s the honest part: fewer people build with Astro today, so finding the next developer is newer ground. But a lot of people are doing it, and the number is growing fast.
And this is the part I think matters most. A well-built Astro site lives in a repository, a secure online home for the code that keeps every past version, with everything documented alongside it. Any developer who points their own AI tools at that repository can have the AI learn how the whole site was built, from the code and the notes. That’s the world we’re moving into.
What to Do Before Your Next Redesign
You don’t need to do anything today. But stop assuming WordPress is the gold standard, or that you need it.
Start with your business instead:
- What does your site actually need to do? Bring in calls and forms, sell online, take bookings, publish every week?
- Who’s going to make changes? Your team, or the person who runs your site?
- Who’s keeping it secure? Every plugin and every login is something someone has to watch.
- What’s still unknown? Astro, hosting on Cloudflare and the tools around them are newer. Ask whoever builds it how they handle updates, backups and handing it off.
If your team is in your site every week publishing content, or your site depends on tools built for WordPress, it may be exactly the right fit. Many of the sites I host still run on it, and plenty should stay there.
Astro Might Not Be the Winner
One more thing, and it matters. This is very early.
Astro might be the name we hear about for a year or two, and then something else comes along that ends up as the real leader 10 years from now. I’ve learned not to rush into things headfirst. When I do adopt something early, it’s because the shift underneath it looks real, not because of the name on it.
So I’m not telling you to switch to Astro. The part I’d bet on is the shift: websites built as custom code with AI, documented and kept in a repository you own. If that’s how your site is built, whatever comes next can pick it up.
It’s Here
Will this end up 60-40 WordPress? 90-10 Astro? Nobody knows. I don’t.
What I do know is that it’s here, it’s real and it isn’t going away. The web is going to look different, and the businesses that understand that early will have the most choices about what comes next.
If you're wondering whether your site should stay on WordPress or move, I'm happy to look at it with you. Let's talk.