AI Spending Is Getting Harder To Predict. CIOs Explain Why

CEOs want more AI. CFOs want predictable costs. And CIOs are stuck in the middle.

It's a problem facing virtually every midmarket IT leader right now—everyone wants AI, but no one's exactly sure what the bill is going to look like. As AI becomes embedded in SaaS platforms, many CIOs are encountering new pricing models that are far less predictable than traditional software licensing.

In the latest episode of MES' Ready.Set.Midmarket! podcast, guests WIN Brands CIO Ken Knapton and Mythworx CEO Jason Williamson discuss how AI is reshaping software pricing models—creating significant cost management and governance challenges for midmarket IT executives.

For years, software-as-a-service costs have been relatively predictable. Organizations licensed software based on users—a simple pricing model that AI is beginning to disrupt. AI is introducing consumption-based pricing, token-based billing, and usage-driven AI services.

As vendors incorporate AI into their existing SaaS offerings, IT leaders may not only find it difficult to forecast AI pricing, but also in justifying its cost.

"The challenge," Knapton said, "is that AI is so new, there's a lot of experimentation going on. We don't really know how much value we're going to get from it yet."

The conversation also drew parallels between today's AI adoption and industry's move to cloud computing roughly a decade ago. Williamson argued that AI is following the same patterns, beginning with experimentation, expansion into production workloads, and then tackling governance and costs-management challenges.

Williamson also spoke about "token sprawl," where employees consume AI resources through multiple tools and platforms without understanding the costs. That lack of awareness can introduce AI-related cost tracking and management complexities.

Knapton weighed in on the importance of governance. At WIN Brands, he said that IT works closely with finance to manage technology spending while also setting policies around approved AI tools and use.

"We've been down this road before," Knapton said, comparing today's AI adoption with the cloud computing era.

For midmarket IT leaders preparing 2027 budgets, the discussion offers a timely reminder that AI adoption is no longer just a technology decision. It is also a budgeting, governance and risk-management challenge.

The full episode can be watched on YouTube and heard on Spotify and Apple Podcasts.

Previous RSM! episodes are here.