Anaconda, the company behind the Python package manager that most data science teams use every day, acquired Kilo Code on July 15, 2026. Kilo is an open-source, model-agnostic AI coding agent with more than three million users across VS Code, JetBrains, the web, and the CLI.

The price was not disclosed. But the problem the deal is solving is easy to put a number on.

The Tokenpocalypse Problem

Kilo Code alone routes nearly ten trillion tokens per month through its platform. That is not a typo. Ten trillion. Across more than 500 models, from frontier providers like Anthropic and OpenAI to open-weight options hosted locally.

That scale tells you something about how fast AI tool adoption has moved inside technical teams. What it does not tell you is who approved those tokens, what they cost, or whether the data being processed was allowed to leave the building.

Anaconda calls the resulting management headache the “Tokenpocalypse.” Most enterprise AI deployments have fragmented across dozens of tools, accounts, and providers, with no consolidated oversight and no easy way to understand what the team is actually spending or doing. Developers find the best tool for the job and use it. Finance finds a five-figure bill and asks questions nobody can answer.

The acquisition is Anaconda’s answer to that problem.

What Kilo Code Brings

Kilo grew from zero to three million developers in sixteen months. That pace is rare, and the reason matters: developers recommended it to other developers because it does not force a choice between your preferred model and your preferred environment.

The platform acts as a model gateway. It routes requests intelligently across whatever models are available, selects the best one for the task at hand, and gives development teams analytics to see what is happening across the entire AI stack. For enterprises, that means:

• A single pane of glass over all AI coding activity, not a different dashboard for every tool
• Policy controls that can enforce which models are approved for which types of data
• Self-hosted options for teams that cannot send code to external APIs
• Freedom to add new models without renegotiating contracts or retraining developers on a different interface

Early enterprise customers using Anaconda’s combined platform are reporting 30 to 50 percent reductions in token consumption, simply from intelligent routing and eliminating redundant calls.

Why This Matters Beyond Data Teams

This acquisition is not just a developer tools story. It is a sign of where enterprise AI governance is heading.

For the past two years, the pattern inside most businesses has been the same: AI tools got adopted faster than policies could be written. Teams started using ChatGPT, then Claude, then Copilot, then Cursor, then Kilo, then whatever launched last month. Each tool had its own billing relationship, its own data handling terms, and its own learning curve.

Now the bills are real. The security teams have questions. And the finance teams want to know what ROI actually looks like across all of it.

Anaconda is betting that the enterprise consolidation layer, the place where all of these tools get governed, governed, and measured, is the most valuable real estate in AI right now. They are probably right.

What This Means for Business

If you are running a data team or deploying AI across any technical function, the Anaconda-Kilo deal is a signal worth watching.

The companies that got the most out of AI in 2024 and 2025 were the ones that moved fast. The companies that will get the most out of AI in 2026 and beyond are the ones that also move smart: with governance in place, costs under control, and a clear picture of what their AI stack is actually doing.

That is not a reason to slow down adoption. It is a reason to build on infrastructure that can scale without turning into chaos.

A model-agnostic foundation, the freedom to use the best tool for each task without vendor lock-in, is the same principle that made Anaconda indispensable for Python development. Bringing that discipline to agentic AI development is a logical next move, and one that most enterprise AI stacks desperately need.

For data professionals, this acquisition also validates something that EDNA’s community has been saying for a while: the skills that matter are not tied to any single model or tool. They are the skills of understanding your data, structuring your problems clearly, and evaluating outputs critically. Those skills work regardless of which model your gateway routes to this week.

---

Enterprise DNA helps business teams build the data skills and AI capabilities they need to make decisions with confidence. Explore learning paths at enterprisedna.co or talk to us about deploying AI across your organisation at Omni by Enterprise DNA.