Why the fear of your AI tool bill might cost more than the bill itself

Many NZ organisations don't yet have an AI spend problem. But we sure have a fear of the bills we might get from our people using these ever evolving tools.

The horror stories reinforce that nervousness. One company burned $500M on Claude in a single month. Uber blew through its entire 2026 AI budget by April. Those stories land hard in organisations that haven't thought carefully about how much they're investing in AI, or how quickly that cost profile can shift - the tech, enterprise capability, and pricing models are all moving faster than most budget cycles. But behind those horror stories sits a common thread: organisations that moved fast without guardrails, usage limits, or governance. That's not where most NZ organisations are. Many are still deciding what tools people should even have access to, and that conservatism comes at a cost.

To be fair, it's getting harder to predict your AI bill. Software used to be simple: a licence fee, monthly or annual, per seat. Predictable. Budgetable. Now the licensing model is shifting. Tools that started as subscriptions are adding usage-based layers on top. Microsoft's Copilot is a good example - what began as a per-seat add-on to Microsoft 365now has Copilot Cowork sitting alongside it, with consumption costs on top of the licence. You pay to access it, and then you pay for what you use. That shift from a fixed cost to an unknown variable cost is what makes you finance BP nervous.

When you hold back access because of what the bill might look like, you're not saving money. You're deferring a capability gap that compounds. AI capability isn't static - every quarter the distance between what these tools can do and what your people understand grows wider. The longer access is delayed, the harder it becomes to close.

The cost trajectory is also moving in your favour. Across the market, AI inference costs dropped 67% year-on-year and are down 99.7%since the first generation of large language models. Open-source competition, better hardware and a genuine price war between providers are structural forces. Over the long term, costs will keep falling. So the fear is pointed ata number that's already declining.

The ROI question also tends to get asked backwards. Reduce headcount with AI and the saving is clean, easy to track, shows up in the P&L. But what if your business moves faster? What if customers get a richer experience? What if you can reach markets you couldn't before? Those outcomes need to be deliberately measured, and most organisations aren't measuring them, so the whole case gets judged on cost alone.

Cost alone always looks bad.

Run the real comparison instead. The cost of AI tool access for an employee - licence plus reasonable usage - versus that same person spending 50% of their week on a manual, repeatable process. These processes exist in every organisation. What does freeing up that capacity actually enable the team to do that they never had time for before?

NZ's productivity problem is real and well-documented - our GDP per hour worked sits well below the OECD average, and the gap has been widening for years. But solving it requires investing in the tools that lift what our people can do. Fear of the bill isn't a productivity strategy.

A 2026 RMIT/Deloitte report put a number on waiting: Australia's AI literacy gap is costing the economy $18.9 billion in foregone growth. Enterprises that moved early are already reporting median cost savings of 35% in their first 12 months. The cost of inaction isn't theoretical. It's compounding.

So in short: yes, set budgets, build guardrails, understand the usage model before you roll something out. That's just good governance. But don't let the fear of a variable cost line be the reason your people don't get access to tools that could change how your organisation performs.

The most expensive AI tool is the one you were too nervous to switch on.

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> 

Ready to move beyond beyond pilots with a clear AI map? To truly realise the power of AI, you need more than just tools, you need strategic alignment and a roadmap for scale. We built aicompass.io to help organisations understand where they are and where they want to be across 32 areas critical to the success of your AI strategy. Whether you need to assess your organisation's AI maturity, align your executive team, or build a measurable path from PoC to production, AI Compass provides the clarity and framework to unleash your AI potential. Check it out at aicompass.io

#AIStrategy #EnterpriseAI #AILeadership #NewZealand#FutureOfWork

‍