Taylor Otwell built Laravel for himself. Millions of developers followed.

Taylor Otwell created Laravel in 2011 as a side project. He built it for himself, mainly. A batteries-included PHP framework—database ORM, authentication, authorization, encryption, job queuing, task scheduling—everything you need for a full-stack web application with strong opinions baked in.

He made no money from it for years. Then in late 2013, he built Laravel Forge, a deployment platform where you bring your own AWS credentials and Forge provisions EC2 servers on your behalf. Within a month, he had thousands of paying customers. "Whoops, product-market fit," he laughs. By January 2015, he was full-time on Laravel. Ten years later, the team is 90 people.

PHP's surprising resilience

PHP gets a bad reputation in Silicon Valley circles. Everyone talks about TypeScript and Python. But PHP powers an enormous amount of the web—it was many developers' first introduction to programming, and the language has evolved dramatically since its rough early days.

Taylor is candid: when he created Laravel, PHP was in a dark place. Node was ascending, other frameworks were popular, people were questioning whether PHP was good for their careers. Laravel was "a breath of fresh air." The language itself has become twice as fast, more type-safe, and more secure since then.

From bootstrapped to Laravel Cloud

The big evolution this year is Laravel Cloud—a fully managed deployment platform that represents the evolution of Forge. Where Forge let you bring your own infrastructure, Laravel Cloud runs everything for you. Autoscaling, preview environments, security.

This is Laravel moving upmarket. Historically, the framework has been popular among indie hackers and SMBs. Now they're selling into enterprises, serving companies that were already running complex self-managed Laravel infrastructure and want to shed that burden.

Taylor raised money in 2024 specifically to build this more ambitious offering. The philosophy: the more successful Laravel Cloud becomes, the more they can invest back into the open source framework.

LLMs are shockingly good at Laravel

One unexpected advantage of being an established framework: LLMs are excellent at writing Laravel code. There's years of training data. Taylor's team has leaned into this by building an MCP server that hooks into Claude Code, allowing the AI to query the Laravel docs, run database migrations, and execute PHP code.

The MCP server is even version-aware. If your project is on Laravel 11, the agent won't use features from Laravel 12. This matters because Laravel ships new patch releases every Tuesday—the AI can access documentation for features released that week.

Taylor admits a "little bit of sadness" about this. He spent years crafting delightful developer experience, and now agents write all the code anyway. But overall, it's a net win. A new generation of developers can learn by prompting, just like Taylor learned by copying and pasting scripts and tweaking them.

The cold start problem for new frameworks

Here's an interesting challenge Taylor raises: if you were launching a new framework today, with no training data, how would you get traction? LLMs are much better at generating code for established frameworks. The rich get richer.

Tailwind CSS, he notes, was "accidentally agent-friendly"—it became the default for LLM-generated frontend code almost by accident. The question is whether anyone can build an "agent-first" framework deliberately, or whether the incumbents will always have the training data advantage.

This interview was conducted at AWS re:Invent 2025.