AI infrastructure costs are already reshaping corporate profitability, but most finance teams still lack the visibility and governance to manage the impact. That’s why we sat down with Eric Mulartrick (Boomi) and Ray Rike (Benchmarkit) to dive into the findings of our joint research with Benchmarkit: The State of AI Cost Governance 2025.
The conversation blended the finance and investment perspective (Ray) with the practitioner view from the trenches (Eric), showing how both sides of the house are struggling with the same blind spots.
Five Biggest Takeaways
1. Forecasting Failure Is Widespread
85% of companies miss AI cost forecasts by more than 10%, and nearly a quarter miss by over 50%. For CFOs, that level of unpredictability isn’t just a budgeting issue, it’s a gross margin risk.
2. Margins Are Already Eroding
84% of companies report AI costs reducing gross margins by at least 6%. For those monetizing AI products, a third are seeing hits of 16% or more. This isn’t hypothetical, t’s already on the P&L.
3. Visibility Blind Spots Run Deep
Only 35% of companies include on-premise infrastructure in AI cost reporting, and just half track LLM API costs, even when those APIs are core to their products. Without full cost-to-serve, finance leaders are flying blind.
4. Hybrid Complexity Is the Default
61% of organizations are running workloads across public cloud, private infrastructure, and third-party services. While hybrid brings flexibility, it fragments billing and makes unified cost governance harder to achieve.
5. Charging for AI Creates Discipline.
Companies that charge for AI are 2–3x more likely to track cost-to-serve precisely, use real-time alerts, and attribute costs to products or customers. Monetization pressure forces the rigor most teams still lack.
What’s Next
As we head into 2026 budget planning, the message is clear: AI costs aren’t just infrastructure, they’re business risks hiding in your margins. Finance, product, and engineering leaders need unified visibility, early-warning controls, and new financial KPIs for the AI era.
If you’d like help assessing your AI cost governance maturity or embedding financial control at the source, contact us to learn how Mavvrik can help.
Thanks to Our Guests
A huge thank you to our speakers for bringing both the practitioner and CFO/VC perspectives to this important discussion. Give them a follow!
- Eric Mulartrick, Boomi
- Ray Rike, Benchmarkit