Plenty of Australian executives will tell you their business is “doing something with AI”. Far fewer can tell you exactly what they are paying for, who they are paying, and whether the invoices arriving each month reflect a considered strategy or a scattering of individual subscriptions that nobody has stitched together. That gap between the story companies tell about artificial intelligence and the money that quietly leaves their accounts is precisely what a new piece of research sets out to measure.
The Weel Australian AI Spending Index, updated for the second quarter of 2026, takes a deliberately unglamorous approach to a heavily hyped subject. Rather than surveying leaders about their intentions, it draws on anonymised expense data from more than 4,000 companies that use Weel’s spend management platform, then tracks how many of them are paying for AI tools and which providers are collecting the money. It is, in effect, a read on AI adoption taken from the corporate card rather than the boardroom.
Why expense data tells a different story
Weel, the Melbourne-founded spend platform previously known as DiviPay, sits in an unusual position to run this kind of analysis. Because its software issues virtual cards and processes business expenses, it can see the transactions themselves, stripped of names and identifying detail, across a large base of Australian firms. Software subscriptions, tool trials and per-seat licences all pass through that pipe. When a marketing manager expenses a writing assistant or a developer signs up for a coding copilot, the charge lands in the same data set as the office coffee order.
That matters because the two dominant ways of measuring AI uptake both have blind spots. Vendor announcements tell you how many seats have been sold globally but reveal little about a single national market. Executive surveys capture ambition and sentiment, which is useful, but ambition is cheap and notoriously prone to overstatement. Expense data captures something harder to fake: a business decided the tool was worth an actual, recurring payment. It also picks up the quiet, bottom-up adoption that never reaches a formal procurement process, the phenomenon usually described as shadow AI, where employees buy tools on their own initiative and the finance team finds out later.
The news in the numbers
The headline finding of the index is directional rather than dramatic: paid AI adoption among Australian businesses is broadening, moving from a cluster of early enthusiasts toward something closer to a standard line item. The Q2 update tracks both the share of companies with at least one AI subscription and the mix of providers taking that spend, giving a rolling picture of a market that has changed faster in eighteen months than most software categories change in a decade.
The provider picture is where the detail gets interesting. A handful of large names, led by the big American model developers and the productivity suites that have folded AI features into existing subscriptions, account for a substantial slice of transactions. Beneath them sits a long tail of specialist tools covering writing, design, transcription, customer support and code. For a chief financial officer, that shape is the whole point. A neat pie chart dominated by one or two vendors suggests a managed rollout. A sprawling list of small, overlapping charges suggests the opposite, and quite possibly a budget leak that nobody has been asked to justify.
Two ways to read it
To the optimists, a rising adoption curve is exactly what a productivity-hungry economy should want to see. Australia has spent much of the past year being told, by the Reserve Bank and by Treasury alike, that weak productivity growth is the country’s central economic problem. If businesses are voluntarily paying for tools that help staff write faster, code faster or handle more customer queries, that is money flowing toward the very efficiency gains policymakers keep asking for. On this reading, the index is a quiet good-news story about grassroots technology adoption running ahead of the official rhetoric.
The more cautious view is that spending is not the same as value. A subscription proves intent, not return. Some of the tools showing up in the data will be delivering genuine gains, while others will be barely used licences that a manager forgot to cancel, or duplicate products bought by three different teams who never compared notes. The risk of AI bill shock, where usage-based pricing produces invoices that balloon without a matching lift in output, is a live concern for finance leaders, and an index built from expense data is as good a place as any to watch that tension play out. Rising spend, in other words, can be a sign of momentum or a sign of drift, and the same chart supports both interpretations depending on what sits behind the numbers.
What it means for Australia
For Australian businesses specifically, the value of a domestic index is that it removes the temptation to assume the local market behaves like San Francisco. Adoption patterns here are shaped by a smaller pool of large enterprises, a vast base of small and medium firms, and a regulatory conversation about AI governance that is still being written. Knowing which tools Australian companies actually pay for, rather than which ones dominate global headlines, gives boards a more honest benchmark for their own behaviour and a way to ask whether they are ahead of the pack or quietly falling behind it.
There is also a governance dimension that lands squarely on Australian desks. Regulators and directors are increasingly expected to understand where corporate data is going, and every AI subscription is a potential channel for sensitive information to leave the building. A finance-led view of AI spending doubles as an early map of that exposure. If the index can show a business how many distinct AI vendors its staff are paying, it has also shown that business how many separate relationships its risk and privacy teams may need to account for, at a moment when data sovereignty and shadow AI are both climbing the enterprise agenda.
What’s next
As a quarterly product, the index is built to be watched over time rather than read once. The most valuable signal will come from the trend line: whether adoption keeps broadening, whether spend concentrates around a few dominant platforms or continues to fragment, and whether the arrival of cheaper open-weight models begins to pull dollars away from the incumbents. Each of those movements would tell a different story about where Australian AI spending is heading and how disciplined businesses are becoming about it.
For now, the contribution is more basic and, arguably, more useful. In a category drowning in projections and vendor optimism, a measure grounded in real transactions offers Australian leaders a rare thing: a way to check the hype against the ledger. The next quarter’s update will show whether the money keeps following the narrative, or starts to tell a story of its own.
Sources: Australian FinTech.
