Learn PHP in 2026 (Yes, Really)
View original- best
Summary (TL;DR)
PHP's reputation lags behind its modern reality. Despite claims of its death, it still powers roughly 43% of the web, mainly through WordPress. The language has evolved significantly since 2010, with PHP 8.x adding enums, readonly classes, and other modern features. The negative reputation stems from its early days of shared hosting and spaghetti code, but the market still demands PHP skills, especially for freelance and agency work. As AI tools generate code across languages, the importance of language choice diminishes, making shipping products more valuable than debating tools.
Learn PHP in 2026 (Yes, Really)

Let's talk about the elephant in the room. No no… not that one. I'm referring to Hypertext Preprocessor, the old grumpy grampy grandpa, PHP.
You have seen the post. It shows up on Hacker News or Reddit every eight to twelve months. Someone writes a long argument about why PHP is finished. The comment section fills with war stories from 2007.
A few hundred upvotes later, the conversation moves on. And then nothing changes.
WordPress still runs roughly 43% of the web. WordPress is PHP. That number has not dropped in a meaningful way for over a decade. It has gone up.
This is a post about reputation. A programming language earns a name for itself. That name sticks after the language changes. This can't be a coincidence.

PHP grew up in a specific place. In the early 2000s, shared hosting was the way most people put a website on the internet. You uploaded files over FTP. You edited code inside a folder called public_html.
There was no build step and no deployment pipeline. You changed a .php file. You refreshed the browser. The change was live. This was fast. It was messy. The language let you mix HTML and SQL in the same file. It did not stop you from writing spaghetti (I'd argue no language does, but that's for another post).
PHP deserved that reputation and the era ended.
However, the reputation did not.
The language a programmer dismisses tells you more about when they formed their opinions than about the language itself.
PHP 5.3 introduced namespaces in 2009. PHP 7.0 doubled performance in 2015. The same release added scalar type declarations. PHP 8.0 brought union types, named arguments, and the match expression.
By 8.1, the language had enums, fibers, and readonly properties. By 8.3, it had typed class constants and deep cloning improvements.
Here is what a value object looks like in PHP 8.3:
enum Currency: string
{
case USD = 'USD';
case EUR = 'EUR';
case GBP = 'GBP';
}readonly class Money
{
public function __construct(
public int $amount,
public Currency $currency,
) {
if ($amount < 0) {
throw new \InvalidArgumentException('Amount cannot be negative');
}
}
public function add(Money $other): self
{
if ($this->currency !== $other->currency) {
throw new \InvalidArgumentException('Currency mismatch');
}
return new self($this->amount + $other->amount, $this->currency);
}
}HOLY f….
The class is immutable. The properties are typed. The enum is backed by a string... These… these features did not exist ten years ago!
The PHP of 2010 and the PHP of today share a name. Not much else.
Programming languages carry social signals. Watch how people talk about their language choices. The signals are hard to miss. Rust signals rigor. Go signals pragmatism at scale. TypeScript signals frontend seriousness. Python signals data fluency or rapid prototyping.
PHP signals nothing a programmer wants to signal. It carries the residue of shared hosting, WordPress theme shops, and Upwork freelancers. The association is blue-collar. The industry pretends class does not exist in programming.
A developer picking a language in 2026 faces two questions. The first is technical: can this language do what I need? The second is social: what will people think of me for choosing it? Yes, really!
The social question wins more often than anyone admits, particularly in a coding interview!
A programmer picking a language to learn is making a career bet and a social bet at the same time. They rarely separate the two.
If the goal is paid work within six months, that narrows the options. WordPress theme development, plugin work, and small business sites make up a large freelance market. The demand is unglamorous and steady.
A junior developer with PHP and WordPress skills can take a paid project within months. The loop from code to payment is short. The React and Next.js path produces better portfolio pieces. Paid client work takes longer to land on that path.
I am describing where the money sits. The market does not care about Hacker News sentiment. Agencies need WordPress developers. Small businesses need someone to edit their theme without breaking their site.
Every language community does this to some degree: JavaScript developers spent years dismissing CSS as “not real programming” and TypeScript as a Micro$oft invasion. Python developers looked down on R. Go developers mocked Java’s verbosity; their own tools generated just as much boilerplate. After Java developers started wearing glasses because they couldn't C#, Rust developers refused to let them borrow a pair.
The pattern is the same each time. A language gets associated with a certain kind of work or a certain era. The association hardens into a judgment. The judgment persists after the original conditions change.
PHP shows this pattern more than any other language. Forty-three percent of the web is a large number. That language gets treated like a punchline at conferences but half the speakers’ personal sites run on WordPress.
Programmers prefer talking about languages over using them. The conversation tracks social norms, not merit.
But that conversation has a shelf life.
LLMs generate working code in PHP, Python, TypeScript, and Go from the same prompt. The developer experience quality gap between languages narrows with every model release. “I specialize in language X” meant something three years ago. It means less today. An employer in a few years will care more about the finished product. The language underneath will matter about as much as the brand of text editor.
Some programmers spent the last decade debating acceptable tools. Other programmers spent it shipping. One group has a portfolio of opinions. The other has a client list and every year spent on taste is a year without invoices.
The McDonald’s application does not ask about your tech stack.
If you liked this, you might like readplace.com, built for exactly this kind of reading.
Thanks for reading. If you have some feedback, reach out to me on LinkedIn, Reddit or by replying to this post.