Open any Q3 board pack right now, and there’s likely a new line item: “AI Cloud Costs – Unplanned Overage”. Sound familiar? Over the last year, I’ve watched multiple leadership teams discover that chasing AI capability through cloud providers often means losing sight of who’s setting the price and whether you’re getting lasting value. Here’s what I wish every board knew before next year’s tech cycle resets.
Most businesses treat AI-as-a-Service as ‘buying convenience’ – until the bills spiral. That’s not a fluke. IBM’s recent industry report found, “The average cost of computing is expected to climb 89% between 2023 and 2025. A staggering 70% of executives IBM surveyed cite generative AI as a critical driver of this increase. And the impact is already being felt across industries, with every executive reporting the cancellation or postponement of at least one generative AI initiative due to cost concerns.”
Every interaction, every agentic automation you deploy, gets billed. If your teams experiment (which you want), you pay for each run, tweak, or accidental retrain cycle. This model suits the provider, not you.
Think about cloud AI as you would a contractor workforce: great for instant results, but expensive, externally controlled, and never embedded in your culture or knowledge base. Once your volume and dependency cross a threshold, you’re funding someone else’s margin and future, not your own.
As Jacob Dencik wrote in the same report, “The cost of computing, often reflected in cloud costs, will be a key issue to consider, as it is potentially a barrier for them to scale AI successfully.” If every new use case tips your OPEX line further into red, scale becomes risk.
For a decade, the usual advice was: “Always buy, never build” — software, not infrastructure. But the economics of AI and fast-cycle, agentic workflows are up-ending that rule. Persistent per-run fees dodge unit cost improvements, so you never see the upside of tech advances unless you own part of the stack.
The old playbook hides two things:
As board stewards, we owe it to our organisations to challenge whether ‘renting intelligence’ can ever deliver the leverage and governance the next five years will demand.
Innovation and budget discipline must go hand-in-hand. As Jacob Dencik cautions, “Even if something is technically feasible to do with AI, if the business case doesn’t stack up because of the cost of computing or the cost of training these models, then we’re not going to see the impact of AI on business activity that many people anticipate.”
What do I hope you take away? This isn’t hype. As of September 2025, the fastest way to lose agility and value is to ignore your real AI cost model. The window to seize control is short – but decisive action now will put you ahead of the herd.
Links:
Quotes: