What doesn’t change:
- BUN remains open-source and MIT-licensed
- The bun is being actively maintained
- The same team still works on the bun
- bun is still publicly available on GitHub
- Bunn’s roadmap will continue to focus on high-performance JavaScript tooling, Node.js compatibility, and replacing Node.js as the default server-side runtime for JavaScript.
Cloud code is sent as executables to millions of users. If the bun is broken, the cloud code is broken. Anthropic has a direct incentive to keep the bun looking great.
what changes:
- We will help make coding tools like Cloud Code and Cloud Agent SDK faster and shorter
- We take a closer look at the first AI coding tool and what makes Bun better for it
- Will ship fast.
About five years ago, I was making a Minecraft-y voxel game in the browser. The codebase grew quite large and it took 45 seconds of iteration cycle time to check if the changes worked. Most of that time was spent waiting for the Next.js dev server to do a hot reload.
It was frustrating, and I got really distracted trying to fix it.
I started porting ESBuild’s JSX and TypeScript transpilers from Go to Zig. Three weeks later, I had a somewhat working JSX and TypeScript transpiler.
Initial benchmarks from the new JavaScript bundler. This transpiles JSX files:
– 3 times faster than esbuild
– 94 times faster than SWC
– 197 times faster than babel pic.twitter.com/NBRt9ESu2d– Jarred Sumner (@jarredsumner) 5 May 2021
I spent most of that first year in a very cramped apartment in Oakland, just coding and tweeting about buns.
runtime
To get Next.js server side rendering to work, we needed a JavaScript runtime. And the JavaScript runtime requires an engine to interpret the code and JIT compilation.
The difference in start time between JavaScriptCore and V8 is interesting. JavaScriptCore appears to start up about 4 times faster.
It’s possible that this is due to the specifics of their respective CLI (rather than JavaScript execution). pic.twitter.com/xd5tSbWf6p
– Jarred Sumner (@jarredsumner) 26 May 2021
So after about a month of reading WebKit’s source code trying to figure out how to embed JavaScriptCore with the same flexibility as Safari, I had an early version of Bun’s JavaScript runtime.
bun v0.1.0
BUN v0.1.0 was released in July 2022. A bundler, a transpiler, a runtime (designed as a drop-in replacement for Node.js), test runner, and a package manager – all in one. We reached 20k GitHub stars in the first week.
Those first two weeks after release were some of the craziest weeks of my life. My job changed from writing code all day to answering people all day. We raised a $7 million seed round led by Kleiner Perkins (thanks to Bucky and Leigh Marie! And thanks to Shrav Mehta, too), I took a salary, and convinced a handful of engineers to move to San Francisco and help build the bun.
bun v1.0.0
Bun started to seem more stable, so we shipped Bun v1.0 in September 2023.
Production usage started to grow and we raised a $19 million Series A led by Khosla Ventures (thanks to Nikita and John!), grew the team to 14 people and got a slightly larger office.
bun v1.1
Even after all this time, we still didn’t have Windows support. And every day, people would ask us the same question: “When will Bun support Windows?”
So we added Windows support and called it bun v1.1. Our Windows support was very difficult at first, but we have made great progress since then.
bun v1.2
Bun v1.2 made major improvements to Node.js compatibility, added a built-in PostgreSQL client and S3 client. We also started to see production use from companies like X and MidJourney. Tailwind’s standalone CLI is built with BUN.
bun v1.3
Bun v1.3 added a built-in frontend dev server, a Redis client, a MySQL client, many improvements bun install And improved Node.js compatibility. Real characteristic: continued growth in production utilization.
AI started getting good
By the end of 2024, AI coding tools go from “cool demo” to “actually useful.” And a ton of them are made with a bun.
Bun’s single-file executable proved to be perfect for distributing CLI tools. You can compile any JavaScript project into a self-contained binary – run anywhere, even if the user doesn’t have bun or node installed. Works with native addons. Fast startup. Easy to distribute.
Cloud Code, FactoryAI, OpenCode, and others are all built with Bunn.
I became crazy about cloud code
I started using Cloud Code myself. I became kind of obsessed with it.
Over the past several months, the GitHub username with the most merged PRs in the bun repo is now Cloud Code Bot. We’ve set it up in our internal Discord and we use it mostly to help fix bugs. This opens a PR with tests that fail in the first system-installed version of bun before they are fixed and passed in the fixed debug build of bun. It responds to review comments. This works perfectly.
It seems like that’s about a few months from where things are going. Certainly not the year.
the way forward
Today, Bun makes $0 in revenue.
One of the most common questions I get asked is about sustainability. Questions like:
“How does a bun become a business?”
“If I bet the bun on my work project or company’s technology stack, will it still be there in five or ten years?”
Our default answer was always some version of “we’ll eventually build a cloud hosting product” that vertically integrates with the Bun runtime and bundler.
But the world of when I first started working on Bun is different from the world of today. AI coding tools are a big change in the way developers do productive work, and the infrastructure layer matters more when agents are writing code.
It felt wrong to force ourselves to follow a set path when AI coding tools were getting so fast, so good.
walk
We have been prioritizing issues for the Cloud Code team for several months. I have lots of ideas all the time and it’s really fun. Many of these ideas also help other AI coding products.
A few weeks ago, I went on a four-hour walk with Boris from the Cloud Code team. We talked about buns. We talked about where AI coding is going. We talked about what Bunn’s team will look like when it joins Anthropic. Then we did this about 3 times over the next few weeks. Then I did the same with many of their competitors. I think Anthropic is going to win.
Betting on Anthropic seemed like a more interesting path. To be at the center of things. Working with a team building the best AI coding products.
it’s a little crazy
At the time of writing, monthly downloads of Bun increased by 25% last month (October, 2025), surpassing 7.2 million monthly downloads. We had over 4 years of runway to figure out monetization. We did not want to be involved in Anthropic.
Instead of putting our users and community through “Bun, VC-backed startup tries to figure out monetization” – thanks to Anthropic, we can skip that chapter entirely and focus on building the best JavaScript tooling.
Why does this make sense?
When people ask “Will Bunn still be around in five or ten years?”, answering “We raised $26 million” is not a good answer. Investors ultimately need returns.
But there is a big question behind this: What does software engineering look like in two to three years?
AI coding tools are getting really good, very fast and they’re using single-file executables of buns to send CLIs and agents running everywhere.
If most of the new code is going to be written, tested, and deployed by AI agents:
- Order And tooling Around that code becomes more important.
- You get much more code overall, written and tested much faster.
- Man is more isolated from each individual line, so the environment in which he lives must be fast and predictable
Bun started with a focus on making developers faster. AI coding tools do the same thing. It’s a natural fit.
Ban joins Anthropic
That’s why we’re joining Anthropic.
Anthropic is investing in Bun as Cloud Code, the Cloud Agent SDK, and the infrastructure powering future AI coding products. Our job is to make Bun the best place to build, run, and test AI-powered software – while remaining a great general-purpose JavaScript runtime, bundler, package manager, and test runner.
Being part of anthropic gives:
- long term stability. A home and resources so that people can safely place their bets on the bun.
- A front row seat to where AI coding tools are, so we can shape the future around that future instead of guessing from the outside.
- More firepower. We are hiring engineers.
And for existing users, the basic promise remains the same:
- BUN is open-source and MIT-licensed.
- The bun is still made in public.
- The same team still works on the bun.
- We’re still passionate about making JavaScript and TypeScript faster to install, build, run, and test.
Anthropic gets a runtime that aligns with software development. We have to work on the most interesting version of that future.
This is going to be really fun.
Why: Is BUN still open-source and MIT-licensed?
A: Yes.
Why: Will bun still be developed publicly on GitHub?
A: Yes. We will still be extremely active on GitHub issues and pull requests.
Why: Does Bun still care about Node.js compatibility and being a drop-in replacement for Node.js?
A: Yes.
Why: Is the same team still working full-time on Bun?
A: Yes. And now we get access to the resources of the world’s leading AI lab instead of a small VC-backed startup making $0 in revenue
Why: What does this mean for Ban’s roadmap?
A: The Bun team will work more closely with the Cloud Code team, and it will likely look similar to the relationship between Google Chrome <> V8, Safari <> JavaScriptCore, Mozilla Firefox <> SpiderMonkey, but with more freedom to prioritize the different ways people and companies use Bun today.
<a href