The CIO Professional Network gathered for a Roundtable discussion led by Kiran Palla on how AI is changing FinOps and what technology leaders need to understand as token-based consumption becomes a larger financial concern. In doing so, Kiran opened the floor for members to share their experience with emerging cost pressures tied to tokens and the difficulty of forecasting AI spend.
Much of the discussion looked at how AI is forcing organizations to rethink FinOps as a viable connector between token usage and business value.
Kiran described tokens as the atomic-level unit behind AI consumption, but emphasized that the real cost picture is shaped by the full AI workflow behind each interaction. Because that workflow can vary widely for each request, user activity alone is not a reliable cost measurement.
The group delved into how leaders can measure AI spend, connect it to business value, manage shadow AI, and prepare for a future where tokenomics becomes a financial metric for the enterprise.
Why It Matters: AI adoption is increasing faster than many organizations can measure or govern its cost. Leaders are encouraging teams to use AI and build AI-enabled workflows, but are discovering that usage is growing faster than budgets and financial controls. However, AI cost management cannot stay confined to engineering. AI’s integration into everyday operations means FinOps needs to account for the full financial picture behind adoption. Without better visibility, organizations risk treating AI as a productivity tool while missing the cost complexity building beneath it.
- Tokens Are Becoming a Financial Metric: Kiran opened by explaining that tokens are now central to understanding AI consumption, but they should not be treated only as an engineering concept. Token consumption is shaped by the full AI interaction leading to tools producing a response. Because that process is often non-linear and hard to predict, user activity alone is not a reliable indicator of cost. He emphasized that the same user action can create very different levels of processing depending on the use case, making tokenomics an important component of enterprise financial planning.
- Context Is a Major Driver of AI Cost: Context window emerged as one of the biggest factors shaping token consumption. Several members agreed that business teams are seeking AI tools that understand the organization more thoroughly because that knowledge can make outputs more useful. However, while these users often understand that value, they likely do not see its effect on token usage, which is creating a serious gap between expectations and financial reality.
- Lower Token Prices Do Not Mean Lower Spend: One leader explained that while per-token prices have become more stable or even lower than before, overall enterprise spend can still rise because consumption is increasing. Agentic workloads and more subsidized usage models have encouraged teams to use AI more often. When these subsidies expire and organizations face real consumption patterns, CIOs, CTOs, and CFOs are going to start to feel the pressure more heavily. Attendees connected this to Jevons paradox, noting that when the price of a commodity falls, usage often rises, making financial discipline even more important.
- AI Cost Is More Than Raw Token Spend: The group also discussed why tokens alone do not capture the full cost of AI. Organizations also need to account for the entire operating model required to support adoption, such as governance and commercial terms. SaaS products can also hide high token-heavy usage costs inside subscription or renewal pricing, making it harder for technology and finance leaders to understand what they are actually paying for. With agentic workloads that span multiple systems and layers, the cost picture becomes harder to attribute unless leaders develop more complete reporting and measurement practices.
- FinOps for AI Requires Normalization and Business Value: FinOps for AI is the operating discipline for translating AI usage into business value. Because of this, the group emphasized the need to normalize usage across vendors and create reporting that can be benchmarked over time. The group compared this to earlier service costing and chargeback models, where technology leaders had to connect infrastructure investment to business delivery. For AI, the same discipline now needs to extend across the full environment so leaders can understand whether that consumption is creating measurable value.
- Business Leaders Need Cost Visibility Without Being Buried in Technical Detail: A participant raised the practical challenge of helping business partners care about consumption when they mainly want solutions that solve business problems. Members responded that organizations need to connect AI features to cost and value at the application level. This approach involves measuring usage over a period of time and then asking business leaders which capabilities are valuable enough to enable or expand. This pivots conversations to discuss whether specific features justify their cost through measurable business value.
- Leaders Need Guardrails for Shadow AI and Future Operating Models: Members noted that employees are facing tension from being pushed to use AI, but also having technology teams tasked with managing budget overruns and inconsistent outputs. Several attendees discussed the need for dashboards and benchmarking models to help teams understand which workloads are worth the investment. That visibility also raised a question about where AI work should run in the first place. This led the group to discuss a potential shift toward smaller internal or purpose-specific models, especially for organizations that want more control over security, compliance, cost, and performance.
Go Deeper – AI Is Changing FinOps. Is Your Organization Ready? (VIDEO) – CION Roundtable
