Published 1 day ago

JetBrains Centralizes AI Spend as Developer Tool Usage Surges 10x in 6 Months

When AI spending rises tenfold in six months, the instinct at many organizations is to cut tools. JetBrains took a different approach: build a control layer underneath the tools instead of removing them.

The company has described how it began centralizing AI usage after development-related spend increased roughly tenfold between January and mid-2026. The trigger was a combination of more developers adopting AI tools and the arrival of increasingly capable frontier models that drove token consumption sharply upward.

4 mins read

6 sections

Key Highlights

JetBrains built a Central CLI to route AI tool requests through a single governed access layer
Managers can now enforce AI budgets at user, team, and org level in real time
The approach preserves developer tool choice while centralizing AI cost visibility and control

The Problem Was Visibility First, Control Second

Most JetBrains developers were using between three and five AI tools each month. That diversity made the bill difficult to read. The company initially spent four days manually collecting usage and expense data into spreadsheets just to get a baseline picture.

Dashboards helped, but they were passive. Seeing where money was going did not provide a way to intervene before it was spent. Requests still traveled directly from individual tools to their respective providers, bypassing any central accounting point.

Central CLI as the Shared Control Layer

The solution grew from an internal tool built by a single developer. JetBrains expanded it into Central CLI, a common invocation layer for both its own and third-party AI tools. Requests routed through the CLI pass through the company’s existing AI platform, creating a single point between developers and model providers.

That shift changed the platform’s role from passive observer to active participant in the traffic generating costs. The practical consequences are concrete:

JetBrains can apply its existing AI-credit system to third-party tools, not just its own.
Managers can view spending and configure limits at the individual, team, or organizational level.
Budget enforcement happens in the flow of requests, not after the invoice arrives.

More than 1,000 developers adopted Central CLI within weeks of its rollout.

Tool Choice Preserved Deliberately

JetBrains was explicit about why it avoided the simpler path of standardizing on one or two approved tools. The argument is straightforward: the pace of change in the AI tools market makes it genuinely difficult to know which tool will perform best for a given task even a few months out. Locking developers into a narrow set today risks locking them out of better options tomorrow.

Centralizing the access layer separates two concerns that are often conflated. Governance and accounting sit in the infrastructure. Tool selection remains with the developers using it.

What Remains Unsolved

The system does not yet cover every source of AI expenditure. Some terminal-based agents and personal subscriptions remain outside the managed path. JetBrains is also still working on policies for distributing AI budgets equitably across different users and teams — a harder problem than simply tracking what is being spent.

The broader challenge JetBrains is navigating is becoming familiar across the industry. The FinOps Foundation has identified generative AI as a growing part of its remit. Other organizations have responded more bluntly: Accenture reportedly asked employees to reduce unnecessary AI use, and Uber introduced monthly limits after exhausting an annual AI budget in four months.

The Practical Takeaway

JetBrains is attempting to solve an infrastructure problem rather than a behavior problem. Rather than asking developers to use fewer tools or less AI, it is building the plumbing that makes consumption measurable and governable at scale.

For teams facing similar cost pressure, the architecture is worth examining: a centralized access layer that routes requests through a shared platform can provide budget enforcement and visibility without requiring a reduction in tool diversity. The tradeoff is implementation complexity and the reality that coverage will be incomplete at first — as JetBrains itself acknowledges.

The approach will not suit every organization. But for those where developer tool choice is a retention and productivity concern, it offers a way to bring AI spend under control without making that control the first thing engineers notice.