JetBrains' Strategy to Control AI Spending
Balancing developer freedom with cost control through innovative solutions
1AI Spending Explosion
JetBrains saw a 10x increase in AI development costs over six months. Developers use 3-5 different AI tools monthly. The release of Claude Opus 4.5 and 4.6 models triggered rapid growth in token consumption. Reaching 150 Claude Code users forced an upgrade to Enterprise pricing. API-based usage costs spiraled out of control in early 2026.
2Management Challenges
Growing demand for diverse AI tools created approval bottlenecks. Multiple sign-offs per request led to ticket backlogs. Wait times hampered productivity. Developers wanted to compare tools like Claude Code, Codex, and GitHub Copilot. Management needed systematic solutions to contain escalating costs while maintaining developer effectiveness.
3Manual Solution Attempts
Initial efforts involved manually downloading data from various consoles and grouping by department. This one-time exercise took four days. The time investment proved unsustainable. Next attempt used APIs to create internal dashboards, but these provided read-only visibility without spending controls. Neither approach offered ongoing cost management capabilities.
4Breakthrough: CLI Wrapper
A developer's personal CLI tool became the solution foundation. It handled JetBrains account authentication and routed requests through the AI traffic layer. Originally built for debugging third-party agents locally, it unexpectedly provided the missing governance functionality. This prototype entered product development in April 2026.
5JetBrains Central CLI
After two months of development, Central CLI emerged with key features: Automated authentication removed login hassles. Auto-detection of installed agents eliminated configuration time. All requests routed through the traffic layer enabled AI credit budgets for third-party tools. The solution met all security, encryption, and access isolation requirements.
6Technical Infrastructure
JetBrains AI Platform has routed billions of LLM requests since 2023. Server-side consumption metrics process through an ETL pipeline into asynchronous, highly elastic storage meeting security standards. Central CLI extended existing credit budgeting to third-party tools while leveraging this proven infrastructure. The solution maintained platform stability throughout implementation.
7Developer Experience
Central CLI lets developers run terminal agents with preferred models without administrative friction. Weeks-long approval waits disappeared. Managers gained visibility through Central Console reports showing current, historical, and forecasted AI consumption by department. Transparent cost tracking replaced blind spending. Productivity remained high while costs came under control.
8Management Benefits
The system enables centralized AI credit budget management. Departments monitor their spending limits. Consumption trends become predictable. Security and compliance requirements are automatically enforced. Cost optimization occurs without sacrificing developer efficiency. Usage patterns inform tool selection and budgeting decisions across the organization.
9Industry Benchmark
JetBrains' experience highlights cost management challenges many companies will face with proliferating AI tools. Centralized solutions become critical for balancing developer freedom with financial control. The CLI-based approach offers an adoption-friendly model for technical teams. This implementation serves as a potential reference architecture for other software companies navigating similar AI cost challenges.
Original article: The JetBrains Blog — JetBrains'ın AI Harcamalarını Kontrol Altına Alma Stratejisi