The Real Cost of Doing Nothing About AI
Doing nothing is also a decision. It just looks invisible on the budget. Three hidden line items that show up later.
Doing nothing is also a decision
Most boards are not yet treating AI tooling as a budget line. That makes "do nothing for now" feel like the prudent option. No new vendor, no new contract, no new capex.
The cost shows up anyway. It just hides on different lines: eventual per-seat licences, Shadow IT incidents, the productivity gap that opens up between teams that use AI and teams that wait. None of these show up neatly in a quarterly review. All three are real.
The point of this article is not to push a purchase decision. It is to argue that "wait and see" is not actually a free option. It has a price. Worth knowing what it is before you choose it.
Three line items that show up later
The eventual per-seat bill
Most "do nothing for now" timelines end with a procurement decision twelve months later, when leadership demands a response to "what is our AI strategy". The default answer is usually a per-seat licence rolled out across the organisation, billed per user per month. For 500 employees at typical enterprise pricing, that is in the order of €180,000 a year. Forever.
The Shadow IT exposure
While leadership decides, employees keep using public AI tools on personal accounts. Confidential data leaves the perimeter daily. There is no audit trail. The first time something becomes a regulator question or a customer trust question, the cost is no longer hypothetical. Both GDPR fines and customer churn from a publicised incident dwarf any platform investment.
The compounding productivity gap
The teams using AI today are getting faster every month. The teams waiting are not. After a year, the gap is visible in throughput, quality, and recruitment. Catching up costs more than starting on time, because you are now training people on tools their peers have used for a year.
Three scenarios, side by side
Stylised numbers and trade-offs for an organisation of around 500 employees. The exact figures depend on size, sector, and how you currently use AI tools.
Do nothing
Hard to measure. Real.
- •Shadow IT data leakage continues
- •No audit trail when regulators ask
- •Productivity gap compounds quietly
- •Eventually forced into a rushed decision
Per-seat licensing for everyone
~ €180K for 500 employees
- •Predictable cost, but high
- •Locked to one model provider
- •Employees use less than half the seats they pay for
- •No department isolation, no dashboard
Sanctioned platform
Flat platform fee + token usage
- •Whole organisation gets access
- •You pay for what people use, not who has a licence
- •Department isolation, audit trail, model freedom included
- •Shadow IT redirected at the network layer
Where these numbers come from
73% Shadow IT figure: Microsoft Work Trend Index, share of employees using AI tools not approved by their employer.
€180K per-seat estimate: 500 employees at €30 per user per month, the typical advertised price for major enterprise AI suites in the European market. Real-world contracts vary.
Productivity gap: Anecdotal but consistent across our customer base. We do not have a clean published study yet. The teams that have been using governed AI for a year are noticeably faster than the teams that are starting now.
The numbers are useful as orders of magnitude, not as exact forecasts. The shape of the argument holds.
Related articles
Ready to see it in action?
Schedule a personalised demo and see how the Plainsight AI Assistant fits your organisation.