How Replit Grew: From 1M MAU to a $9B Valuation

Replit grew in two very different stages.

First, it made coding easier to start and easier to share. The product ran in a browser, teachers brought entire classrooms onto it, students shared projects with friends, and YouTube tutorials sent new developers directly into Replit.

That built a huge audience. Replit had more than 20 million developers by early 2023, but revenue remained small.

Then Replit Agent changed the business.

Instead of helping programmers write code, Agent let almost anyone describe an app and have Replit build, run, and deploy it. Replit says annualized revenue grew from $2.8 million to $150 million in less than a year after Agent took off.

In March 2026, Replit raised $400 million at a $9 billion valuation. By September 2026, the company said it had more than 60 million users worldwide.

The interesting part is that Agent was not an overnight success.

Replit spent more than a decade building the pieces that made it possible.

Replit's growth in numbers

Year	Milestone
2011	Amjad Masad launches an early browser-based REPL
2018	Replit reaches about 1 million monthly active users
2021	Registered users pass 6 million
2022	Replit reports 15 million users
2023	Replit reaches 20 million developers and 240M+ Repls
2024	Replit Agent launches
2025	Replit reports 40M+ users and $150M in annualized revenue
Mar. 2026	Replit raises $400M at a $9B valuation
Sep. 2026	Replit reports 60M+ users worldwide

The growth was not driven by one channel.

It came from several loops that built on each other.

1. Replit removed the setup required to start coding

Replit started with a simple problem.

Amjad Masad wanted to code from different computers without setting up a development environment every time.

Normally, starting a programming project meant installing a language, configuring packages, opening a terminal, and making sure everything worked on your machine.

Replit moved that work into the browser.

You could open a page, start writing code, and run it.

That made the product especially useful for beginners. A student did not need to understand how to configure Python before writing their first Python program.

But removing setup did something else too.

It made code easy to share.

Instead of sending someone files and instructions for running them, a Replit user could send a link.

The other person could open it immediately.

That link became one of Replit's earliest growth features.

Teachers could send exercises to students. Developers could share examples. Students could show games to friends. Someone could discover a project and start building their own version.

The product was easier to distribute because the product itself lived on the web.

2. Teachers brought entire classrooms onto Replit

Education became one of Replit's first big growth engines.

The reason was simple: Replit did not have to acquire students one at a time.

It could acquire a teacher.

Former Replit VP of Growth Patrick Coleman said that each teacher who signed up could bring about 25 students with them.

Those students then shared Replit with friends.

So the loop looked something like this:

teacher discovers Replit → teacher invites a class → 25 students start using it → students build projects → students share those projects → more people discover Replit

That is very different from paying for 25 individual signups.

Replit started selling its classroom product in 2017 and reached roughly $150,000 in annualized revenue within months.

During COVID, the distribution became even stronger.

Schools moved online. Replit already worked in the browser, required little setup, and allowed people to collaborate remotely.

Registered users passed six million by early 2021.

Education was good for user growth.

It was not necessarily a great business.

Schools were price-sensitive. Sales and procurement could be slow. Supporting free compute was expensive. Student accounts also attracted abuse from people using the platform for crypto mining and other activities that had little to do with learning to code.

Replit eventually moved away from education as a dedicated product.

But by then, classrooms had already introduced millions of people to Replit.

3. Sharing a Repl created a built-in viral loop

Education worked partly because Replit projects were naturally shareable.

A project on a traditional development environment usually stayed on the developer's computer until they decided to deploy it somewhere.

A Repl already existed online.

Users could send a link to something they had built. The recipient could run it without installing anything.

That gave Replit a product-led growth loop:

build → share → open → remix → build → share

This worked especially well for young developers.

People like showing other people things they made.

A game was not only a project. It was something you could send to your friends.

A programming experiment could become a tutorial.

A small website could be shared publicly.

A useful tool could spread far outside Replit itself.

By July 2022, Replit said it had more than 15 million users, more than 20 million websites and apps created and hosted, and more than 10 billion monthly visits across Replit-hosted sites.

By February 2023, the company said 20 million developers had created more than 240 million Repls.

Replit was not just acquiring users.

Its users were creating things that brought more people back to Replit.

4. YouTube became one of Replit's best acquisition channels

Replit also found a channel that matched the product almost perfectly: YouTube.

The company did not just make videos explaining why people should use Replit.

It worked with creators who taught people how to build things.

One of its best-performing partnerships was with freeCodeCamp.

Instead of making a video called:

"Why you should use Replit"

the content taught viewers something they already wanted to do, such as:

"How to code a Discord bot in Python."

Replit was simply the easiest place to follow along.

That distinction mattered.

Someone watching a tutorial about building a Discord bot already had intent. They wanted to code something.

Replit gave them an environment where they could start immediately.

Patrick Coleman said traffic coming from YouTube converted to Replit signups at roughly 35%. He also said YouTube and people sharing Repls were far better at bringing in new users than Twitter.

Twitter still mattered for the company. It helped with brand, hiring, fundraising, and engaging existing users.

But it was not nearly as good at acquiring new users.

Replit doubled down on the channel that worked.

It partnered with programming creators, watched which projects people wanted to build, and created more content around those use cases.

That created another loop:

people search for how to build something → find a tutorial → tutorial uses Replit → start building on Replit → share the finished project

The lesson was simple.

Replit marketed the thing the user wanted to make, not the coding environment itself.

5. Multiplayer made Replit more useful and later helped Agent

Replit launched multiplayer in 2018.

Several people could work inside the same project at the same time.

They shared the editor, filesystem, terminal, output, and other parts of the development environment.

It was closer to Google Docs for programming than traditional screen sharing.

This helped growth immediately.

Students could work on projects together.

Teachers could enter a student's project and help.

Developers could debug with coworkers.

But multiplayer became more important years later.

An AI coding agent needs to do many of the same things a human collaborator does.

It needs to inspect files.

It needs to change code.

It needs to run commands.

It needs to see what happened.

It needs to notice errors and try again.

Replit had already spent years building a workspace where another participant could interact with a running software project.

Eventually, that participant could be an AI agent.

6. Replit did more than generate code

Another old Replit decision became important once AI coding took off.

Replit ran the code too.

The company had built infrastructure for languages, packages, containers, databases, hosting, and deployment.

A normal coding assistant could suggest code.

Replit could help write the code and then run it.

That sounds like a small difference.

It is not.

Writing code is only one part of shipping software.

A real application may also need dependencies, databases, environment variables, servers, hosting, authentication, and deployment.

Replit had been building these pieces for years.

The infrastructure was expensive.

Free compute attracted abuse. The company dealt with crypto miners, bandwidth theft, attacks, reliability problems, and rising infrastructure costs.

But the same infrastructure later gave Replit Agent somewhere to work.

Once AI models became good enough to make larger code changes, Replit already had an environment where the model could write, run, test, debug, and deploy the result.

7. Replit had a huge audience but struggled to monetize it

By the early 2020s, Replit had solved distribution much better than monetization.

Millions of people used the product.

A lot of them were students, beginners, hobbyists, or developers using the free tier.

That was great for growth.

It was harder to turn into a large software business.

Professional developers also had strong free alternatives such as VS Code and other local development tools.

By early 2024, Replit's revenue was reportedly still around $2.5 million to $2.8 million a year despite its huge registered user base.

That created an unusual situation:

Replit had distribution.

It had infrastructure.

It had brand awareness.

It had millions of accounts.

It just did not yet have a product that could monetize those assets at the scale the company wanted.

AI changed that.

8. Ghostwriter showed that people would pay for AI

Replit launched Ghostwriter in 2022.

It could complete code, generate code, explain code, transform it, and help developers search through projects.

The important part was not that Ghostwriter suddenly caused Replit's audience to explode.

There is little public evidence that it did.

Replit had already reported 15 million users before Ghostwriter's beta. It crossed 20 million developers a few months later.

Ghostwriter mattered for another reason.

It taught Replit how to put AI inside the development environment.

The company learned how to serve models, reduce delays, give the model project context, and charge separately for AI.

More than 1,000 people were reportedly paying for Ghostwriter by the end of 2022.

That proved some users would pay Replit for AI.

But Ghostwriter still assumed the user knew how to code.

You told the AI what code you wanted.

Agent would later remove that assumption.

9. Replit tried agents before the models were ready

Replit did not suddenly discover agents in 2024.

The company had experimented with autonomous coding during earlier waves of AI.

The results were not good enough.

Models forgot context.

They made poor plans.

They became stuck.

They struggled to make a long series of changes without breaking the project.

Replit had built much of the environment an agent needed.

The models were the missing piece.

That started changing in 2024 as coding models became better at working across larger codebases and completing longer tasks.

Replit reorganized around the opportunity.

The company cut projects.

Its team became much smaller.

CEO Amjad Masad created an internal push around Agent.

The first version was still unreliable.

Replit launched it anyway.

That decision was risky, but Replit had one advantage most new AI coding startups did not have:

years of infrastructure and millions of existing users who could try it.

10. Replit Agent changed the first session

Replit Agent launched in early access in September 2024.

Before Agent, a new Replit user still needed to think like a programmer.

What language should I use?

What packages do I need?

How should I structure the project?

Where should I put the database?

How do I deploy it?

Agent changed the starting point.

The user could simply describe the product.

For example:

"Build me an app that tracks customer feedback and groups similar requests."

Agent could then create the environment, write code, install dependencies, run the project, fix errors, and help deploy the application.

The user no longer needed to start with code.

They could start with an idea.

That changed Replit's potential market.

The old market was:

people who want to code.

The new market became:

people who want software.

That included founders, designers, marketers, operators, students, small businesses, and people who had never written code before.

This was the real AI breakthrough for Replit.

11. Agent made Replit easy to demonstrate

Agent also had another growth advantage.

The product was visual.

Someone could write a short prompt and then show an application appearing on screen.

That made Replit easy to demonstrate on X, YouTube, TikTok, and other social platforms.

The output was also shareable.

Someone could build an app, deploy it, and send the result to other people.

So the old Replit growth loop survived in a new form:

describe idea → Agent builds app → deploy → share → someone sees what is possible → they build their own app

The main difference was that the person entering the loop no longer needed to be a developer.

AI made Replit's old distribution model available to a much larger audience.

12. Agent improved the metric that mattered: getting an app live

Replit's own experiments give a clue about why Agent changed the business.

The company reported that with Agent V2, new users became about twice as likely to deploy an application. Power users became five to six times as likely to deploy.

Replit also found that users who deployed applications were around ten times more likely to retain.

That does not prove deployment alone caused the higher retention.

People who are more engaged may simply be more likely to deploy.

But the direction is important.

The value of Replit was not generating a pile of code.

It was getting someone from an idea to working software.

Agent shortened that path.

That is what made the product useful to people who would never have used a normal coding assistant.

13. Revenue went from $2.8M to $150M annualized in less than a year

The effect on Replit's business was dramatic.

In September 2025, Replit said annualized revenue had increased from $2.8 million to $150 million in less than a year.

That is more than 50x growth.

The company also said it had more than 40 million users.

The same funding round valued Replit at $3 billion.

Six months later, the valuation tripled again.

In March 2026, Replit raised $400 million at a $9 billion valuation.

By September 2026, Replit said it had more than 60 million users worldwide.

Agent did not create Replit's first audience.

Replit already had tens of millions of users.

Agent changed how valuable each new and existing user could become.

It also opened Replit to people who would never have considered themselves developers.

14. Replit turned employee adoption into enterprise growth

After Agent worked with individuals, Replit started pushing further into companies.

This created another version of the classroom loop.

One employee could discover Replit.

They could build a useful internal app.

A coworker or manager could see it.

Other employees could start building.

The company could then become an enterprise customer.

Replit supported this with company hackathons.

Instead of asking a business to sit through a software demo, employees spent a few hours building tools themselves.

At Rokt, Replit says 700 employees created 135 applications in 24 hours.

Replit has also said companies such as Zillow and Meta saw employee adoption before asking about enterprise plans.

The loop looks familiar:

one organizer → group of users → users build → useful projects spread → larger account

It is basically the old classroom loop with a much larger contract at the end.

What did not work for Replit

Replit's growth story was not a straight line.

Education brought millions of users but was difficult to monetize. The company eventually shut down its dedicated education product, which upset teachers who had built classes around it.

The huge free tier created expensive infrastructure problems.

Replit also spent years working on products that did not produce a major revenue breakthrough.

Agent introduced another problem: AI actions cost money, and those costs can be difficult for users to predict. Replit changed its pricing several times and had to refund some customers after a billing error in 2025.

More autonomy also created more risk. In one widely reported 2025 incident, Agent deleted data from a database during a test with SaaStr founder Jason Lemkin. Replit later added stronger safeguards.

These problems matter because Replit is not selling autocomplete anymore.

It is giving software an agent that can take actions inside a real computing environment.

That is far more useful.

It also makes mistakes more expensive.

What actually drove Replit's growth?

Replit did not grow because of one viral launch.

Its growth came from several advantages stacking on top of each other.

The browser removed setup.

Shareable Repls turned projects into distribution.

Teachers acquired groups of students at once.

Students shared what they built.

YouTube captured people who already wanted to build specific projects.

Multiplayer created a collaborative workspace.

Cloud infrastructure let Replit run and deploy the code.

Ghostwriter taught the company how to put AI inside that environment.

Then Agent combined all of those pieces into a much simpler promise:

tell Replit what you want to build, and Replit will help build it.

The company spent years acquiring people who wanted to learn how to make software.

AI let it serve everyone who simply wanted software made.

That is the biggest change in Replit's growth story.

Replit's growth timeline

2011: The browser REPL

Masad launches an early browser-based programming environment and shares it on Hacker News.

The core idea is simple: code should run anywhere without local setup.

2016: Replit becomes a company

Masad returns to the project and starts building Replit as a company with Haya Odeh and Faris Masad.

2017–2018: Education becomes a growth engine

Replit builds classroom tools.

Teachers introduce groups of students to the product.

Replit reaches roughly one million monthly active users by 2018.

2018: Multiplayer launches

Replit makes collaborative programming possible inside the browser.

The technology later becomes useful for AI agents.

2020–2021: COVID accelerates classroom adoption

Remote learning increases demand for browser-based coding.

Registered users pass six million.

2022: Replit reaches 15 million users

The company reports more than 20 million websites and apps built and hosted on Replit.

YouTube, educational creators, and viral Repl sharing become important acquisition channels.

2022: Ghostwriter launches

Replit begins charging separately for AI coding features.

2023: Replit crosses 20 million developers

Users have created more than 240 million Repls.

Replit increasingly combines coding, AI, hosting, and deployment in one product.

2024: Replit focuses on Agent

The company cuts projects and concentrates more of the team on autonomous software creation.

September 2024: Replit Agent launches

Users can describe an application in plain language and have Agent build and deploy it.

2025: Revenue takes off

Replit says annualized revenue grows from $2.8 million to $150 million in less than a year.

It raises $250 million at a $3 billion valuation.

March 2026: Replit reaches a $9 billion valuation

Replit raises another $400 million.

The company reports more than 50 million users and says people at 85% of Fortune 500 companies have used Replit.

That does not mean 85% of the Fortune 500 are paying enterprise customers.

September 2026: Replit passes 60 million users

Replit says it now has more than 60 million users worldwide as it expands internationally.

The biggest lesson from Replit's growth

Replit looks like an overnight AI success if you start the story in 2024.

It looks very different if you start in 2011.

The browser editor mattered.

The classroom product mattered.

The students mattered.

The YouTube videos mattered.

Multiplayer mattered.

Hosting mattered.

Deployment mattered.

Ghostwriter mattered.

Some of those products never became huge businesses on their own.

Together, they created the foundation for Agent.

That is what makes Replit's story useful.

A company's old products, users, infrastructure, content, and distribution can become much more valuable when a new technology changes what they can be used for.

Replit already had millions of users and almost everything needed to build and run software in the cloud.

AI removed the final barrier.

The user no longer had to know how to code.

Frequently asked questions

How did Replit grow?

Replit first grew by making coding easy to start and share in a browser. Teachers brought classrooms onto the product, students shared their projects, and YouTube tutorials brought in high-intent users. Replit later used its existing cloud development infrastructure to launch Replit Agent, which expanded the product beyond programmers and drove a large increase in revenue.

How many users does Replit have?

As of September 2026, Replit says it has more than 60 million users worldwide.

Older public milestones include roughly one million monthly active users in 2018, 15 million users in 2022, 20 million developers in early 2023, and more than 40 million users in September 2025.

These figures are not all directly comparable. The 2018 figure measured monthly active users, while most later figures describe registered or cumulative users.

How did Replit get its first users?

Replit's early users came from the programming and education communities. Its browser-based REPL was shared on Hacker News, used in coding education, and adopted by teachers and students.

The fact that projects could be opened through a link made Replit naturally easy to share.

Was education important to Replit's growth?

Yes.

Education was one of Replit's strongest early distribution channels. Former VP of Growth Patrick Coleman said one teacher could bring about 25 students onto Replit, and students would then recommend the product to friends.

The education business itself was difficult to monetize, but it helped Replit build a very large user base.

How did YouTube help Replit grow?

Replit worked with programming educators and creators to make tutorials around things people wanted to build.

Former Replit VP of Growth Patrick Coleman said traffic from YouTube converted to signup at roughly 35%, and that YouTube and viral Repl sharing were much better at acquiring new users than Twitter.

One successful example was freeCodeCamp content showing people how to build Discord bots using Replit.

When did Replit Agent launch?

Replit Agent launched in early access in September 2024.

It allowed users to describe an application in natural language while Agent handled tasks such as setting up the development environment, installing packages, writing code, running the project, and helping deploy it.

Did Replit Agent create Replit's audience?

No.

Replit already had tens of millions of registered users before Agent launched.

Agent did two things differently: it gave Replit a much stronger way to monetize its existing audience, and it expanded the market from people who knew how to code to people who simply wanted to build software.

How much revenue does Replit make?

The cleanest company-reported growth milestone is from September 2025, when Replit said annualized revenue had grown from $2.8 million to $150 million in less than a year.

Later figures have included higher founder claims and outside estimates, but they do not all use the same definition.

Replit has also said it aims to reach a $1 billion revenue run rate. That should be treated as a target unless the company confirms it has been reached.

What is Replit worth?

Replit raised $400 million at a $9 billion valuation in March 2026.

Six months earlier, a September 2025 funding round had valued the company at $3 billion.

Why did Replit Agent grow so quickly?

Agent launched into a product that already had millions of users, a browser IDE, multiplayer collaboration, cloud compute, packages, databases, hosting, and deployment.

Replit did not have to build all of those pieces after the AI breakthrough.

It could use them immediately.

That allowed Agent to do more than generate code. It could help take an idea all the way to a running application.

Sources

• Original 2011 Hacker News launch, Hacker News
• Amjad Masad on Replit's early growth, First Round
• Replit multiplayer, Replit
• Teams for Education shutdown, The Register
• Ghostwriter launch, Replit
• Amjad Masad on building Agent, Sequoia
• Replit Agent launch, Replit
• Replit's interview with former VP of Growth Patrick Coleman, Compound Manual
• Replit's 20 million developer milestone, Replit
• Replit's September 2025 funding announcement, Replit
• Replit's March 2026 $9 billion funding announcement, Replit
• Replit's September 2026 international expansion announcement, Replit
• Replit reaches a $3B valuation, TechCrunch
• Replit reaches a $9B valuation, TechCrunch