Claude Code has reached $1 billion in run-rate revenue just six months after going generally available, and on the same day Anthropic confirmed it has acquired Bun, the fast JavaScript runtime. Anthropic announced both on 3 December 2025, framing the Bun deal as a way to make its coding agent faster and more reliable. For UK developers and engineering teams weighing AI tooling, this is a clearer signal than any benchmark: the money and the infrastructure are now following the agent.
- Claude Code hit $1bn (about £790m) in run-rate revenue in November 2025, six months after launching in May 2025.
- Anthropic has acquired Bun, the all-in-one JavaScript runtime, package manager, bundler and test runner founded by Jarred Sumner in 2021.
- Bun will remain open source and MIT-licensed, with over 82,000 GitHub stars and more than 7 million monthly downloads.
- It follows Anthropic’s Stainless (18 May 2026) and Vercept (25 February 2026) deals, a clear developer-tooling push.
What the $1bn run-rate milestone actually tells UK teams
Run-rate revenue is an annualised projection from a recent period, not a banked figure, so the honest reading is that Claude Code was earning at a pace that, if held for twelve months, would total $1bn. Anthropic says it cleared that line in November 2025, six months after the tool became publicly available in May 2025. For context on the parent company’s finances, our report on Anthropic’s Q2 2026 profit shows the agent business is no longer a side project.
The practical takeaway for a UK engineering lead is that Claude Code is now a funded, supported product with the revenue to justify continued investment, rather than an experiment that might be quietly retired. That matters when you are deciding whether to standardise a team of ten or fifty developers on a tool. We would still treat run-rate figures with care, but the direction of travel is not in doubt.
Why buying Bun changes the day job for UK web developers
Bun is the part of this story most British web developers will feel directly. It bundles a JavaScript runtime, package manager, bundler and test runner into one tool, competing with Node.js and the npm toolchain that most front-end and full-stack teams in the UK run on daily. Jarred Sumner started Bun in 2021, and Anthropic says it will stay open source and MIT-licensed, which removes the obvious fear that a corporate owner locks it behind a paywall.
Mike Krieger, Anthropic’s Chief Product Officer, said Bun “represents exactly the kind of technical excellence we want to bring into Anthropic” and called it essential infrastructure for AI-led software engineering. In plain terms, a faster runtime under the hood should mean Claude Code installs dependencies, runs tests and rebuilds projects more quickly, which is exactly where an agent loses time today. If your team already runs Bun in production, the upside is direct; if you are on Node, nothing breaks, because the project stays open.
Why does the runtime matter so much for an agent specifically? When Claude Code works through a task it does not just write code once; it edits files, installs packages, runs the test suite, reads the failures and tries again, often looping several times before a change is ready for review. Each of those steps spends real wall-clock time, and on a large UK codebase the install-and-test cycle can dominate the loop. Bun’s pitch has always been that it collapses the separate Node, npm, bundler and test tools into one fast binary, so shaving seconds off every iteration compounds across a working day. For a British team paying engineers by the hour, faster feedback is not a vanity metric; it is the difference between an agent that feels usable on a real project and one that is left waiting on the toolchain.
There is a portability caveat worth stating plainly. Bun aims for broad Node.js compatibility, but it is not a perfect drop-in, and some native modules or edge-runtime quirks still behave differently. A UK team should not assume an existing Node service will run unchanged on Bun simply because Anthropic now owns the project. The sensible reading is that the acquisition makes Bun a more strategic dependency for Claude Code itself, while your own decision to adopt Bun in production should still rest on testing it against your actual workloads rather than on the deal headline.
Stainless and Vercept: the wider developer-tooling buying spree
Bun is the third developer-infrastructure acquisition Anthropic has confirmed in recent months, and the pattern is the point. On 18 May 2026 it bought Stainless, founded in 2022, which generates SDKs, command-line tools and MCP servers from API specifications across TypeScript, Python, Go, Java and more. Stainless has powered every official Anthropic SDK since the API’s earliest days, so the buy formalises a dependency that already existed.
Earlier, on 25 February 2026, Anthropic acquired Vercept, whose founders Kiana Ehsani, Luca Weihs and Ross Girshick worked on letting AI see and act inside the software people use. That feeds Claude’s computer-use ability: Anthropic cites Claude Sonnet 4.6 improving on the OSWorld benchmark from under 15% in late 2024 to 72.5%. Taken together, Stainless covers how agents connect to tools, Vercept covers how they operate software, and Bun covers how fast the code underneath runs. If you are still deciding between assistants, our guide on choosing Claude, Copilot or Gemini for UK work is a useful starting point.
How Claude Code compares with GitHub Copilot and Cursor for UK teams
For a UK team picking an AI coding tool, the three names on most shortlists are Claude Code, GitHub Copilot and Cursor. Claude Code is a terminal-first agent that reads, edits and runs across a whole codebase; Copilot lives inside VS Code and increasingly across GitHub’s platform; Cursor is an AI-native editor built on the VS Code base. The Bun acquisition is Claude Code’s bet that owning the runtime makes the agent loop quicker, where Copilot leans on GitHub’s ecosystem and Cursor on editor ergonomics.
The model underneath matters as much as the harness around it, and this is where Claude Code’s pitch sharpens for British teams. The same Claude models that power the agent also sit behind Anthropic’s API, so a UK shop that already calls Claude for product features gets a consistent reasoning engine across both its app and its tooling, with one billing relationship and one set of data terms to negotiate. Copilot ties you to OpenAI models through GitHub, while Cursor lets you mix providers but leaves the integration work to you. For a lead trying to keep the stack simple, a single vendor for both the coding agent and the production model is a real, if unglamorous, advantage.
Cost and governance matter as much as raw capability for British teams. Anthropic sells Claude through Pro at $20 a month (about £16) and Max from $100 a month (about £79), with the exact sterling figures and what each tier unlocks set out in our Claude UK pricing breakdown. Security teams should also note recent supply-chain risk in dev tooling, as our coverage of the GitHub VS Code extension breach showed; whichever agent you pick, repository access controls need the same scrutiny.
On the migration question, the costs that bite a UK SME are rarely the licence fees and almost always the change in working habits. Moving a team onto an agentic tool means new review norms, because a human still has to read and approve what the agent produces, and it means agreeing where the tool is allowed to run, which repositories it can touch and how secrets are handled. Budget time for that adjustment rather than expecting an instant productivity jump in week one. Sterling figures move with the exchange rate, so treat the roughly £16 and £79 monthly equivalents as indicative and confirm the live rate and any VAT on the invoice before you commit a department.
For most UK SMEs the realistic move is a small pilot on one repository before any team-wide rollout. Track time saved per pull request, not headline benchmarks, and check that your data-handling matches the plan you buy. Our look at the Anthropic Economic Index and at the Claude Opus 4.8 developer upgrade both add useful UK context on how coding work is shifting.
Where UK developers and engineering leads should check next
Start with the source: read Anthropic’s own announcement for the milestone and the Bun deal, then check the Bun GitHub repository to confirm the project’s open licence and release cadence have not changed. For procurement, price Claude against your current spend on GitHub Copilot Business or Cursor seats, and ask each vendor where UK customer data is processed and stored before you commit a team. If you run regulated workloads, our piece on enterprise agent rollouts covers the governance questions worth asking.
Engineering leads should also run a short internal trial against a real backlog item rather than a toy demo, because agent quality varies hugely by codebase. Measure review time, defect rate and how often a human has to intervene. Those three numbers will tell a UK team far more than a $1bn headline ever can.
| Tool | Best fit | MTW read |
|---|---|---|
| Claude Code | Whole-codebase agent work from the terminal | Bun buy targets a faster agent loop; strongest momentum right now |
| GitHub Copilot | Teams already deep in GitHub and VS Code | Best platform integration; safest default for GitHub-centric shops |
| Cursor | Developers who want an AI-native editor | Smoothest editing UX; weigh the per-seat cost for larger teams |
Our verdict
Our view is that the Bun acquisition matters more to UK developers than the $1bn run-rate headline, because it is a concrete bet on the infrastructure under the agent rather than a marketing number. If your team already writes JavaScript or TypeScript and is curious about agentic coding, this is a sensible moment to run a paid pilot of Claude Code on a single repository, with the open-source Bun runtime giving reassurance that the tooling stays portable. Teams locked into the GitHub stack should stay with Copilot for now and watch how quickly the Bun-powered speed gains land. We would wait before any team-wide migration until Anthropic publishes UK data-processing detail and you have your own time-saved numbers from a real backlog. The risk that would flip our call is Bun’s open licence or release pace changing after the deal; if that happened, the portability argument weakens and the maths shifts back toward the incumbents.
Reader discussion
Leave a comment
Comments are moderated. Keep it useful, accurate, and on topic.