PLANNING FOR THE AGE OF AI TOKENS
The Industrial CFO's Guide to Planning and Budgeting for AI
A perspective for C-suite, CFOs, and finance leaders in industrial companies
Industrial companies face a familiar-looking decision: how much to spend on a rapidly evolving technology. The difficulty is that its cost structure behaves nothing like the ones their planning processes were built for.
AI spend is usage-driven rather than seat-driven. It moves in orders of magnitude rather than percentages. And it is scattered across cloud bills, SaaS features, and shadow purchases. Mature teams forecast cloud spend within 1–3% and miss AI spend by a factor of two to three.¹ This is a different discipline, not a bigger version of the old one.
This whitepaper sets out how CFOs and finance leaders at industrial companies can plan, fund, and monitor AI spend when the cost behaves like a metered utility, and how to keep adoption growing without letting the bill run wild. It draws on FinOps Foundation research, industry surveys, and Aicadium's work with industrial clients.
What’s Inside:
The Four V's problem with AI costs
Visibility, Variability, Volatility, and Value: why AI cost does not behave like per-seat software, and why industrial companies are building "FinOps for AI" rather than folding AI into the IT budget.
Two portfolios, two governance models
Improving what you already do has a "before" state and is budgeted like other software. Building a capability you do not yet have is funded like a venture portfolio: staged funding, a six-question scorecard, and kill criteria agreed before the pilot starts.
How to monitor returns
Three gates, leading indicators before financial ones, and cost per outcome as the standard unit metric, with the baseline owned by Finance and captured before go-live.
Promoting AI usage without letting costs run wild
Tiered access, showbacks and chargebacks, budget tied to value rather than headcount, and the engineering-side levers to pull before reaching for a cap. Closes with ten actions for CFOs.