In June of this year, for the first time, the subscription software spend of companies on our platform split almost evenly between AI tools and traditional SaaS.
All of us have seen headlines about companies like Meta and Amazon capping token budgets after employees racked up spend way beyond expectations. While Meta and Amazon can easily shrug off a billion dollar receipt, that’s not true of the vast majority of companies.
)
Consolidate and connect: Your stack needs to be in one place to progress
When we interrogated our data, we expected to see total tool count increase with time. Instead, over the past 12 months, it barely moved: just 5% growth in total number of tool subscriptions. But look at what happened inside that 5%: Non-AI tools spend peaked and started contracting. AI tools jumped 72% in the last 12 months.
)
Companies are investing deeply in fewer, higher-value platforms — and those platforms are AI. AI's share of total SaaS spend went from 2% to 9% in a single year. That is more than 4× increase, while companies added less than half an AI tool per company, on average.
The SaaS stack is shifting. We're in Phase 1 right now: AI adoption and expansion. Phase 2 will be about connecting those strategic AI tools directly to the data they need because no matter how advanced a model is, it’s still only as intelligent as the context its user provides.
Companies will want one consolidated spend layer feeding real data directly into one AI assistant, which becomes their workstation. They’ll need a bridge from Spendesk's data directly into the tools they're already investing in.
We know because we saw it happen. A few months ago, our CFO Pauline stopped opening Spendesk for her reporting queries. She'd moved to her AI assistant. But she was uploading reports manually, exporting data every time she had a question. She hadn't found a better tool. She'd found a faster interface, but it was still disconnected and manual.
The bottleneck, she realised, was the lack of connection between the AI and her live spend data.
That's why earlier this week, we launched Spendesk AI Connect. It connects the spend data companies have already consolidated in Spendesk directly to Claude, Dust, and other AI assistants of their choice, removing the manual work that still separates AI from live financial data. Finance teams get faster access to answers, while AI gets the context it needs to move one step closer to true agency.
Spendesk AI Connect launched this week. Read more about how it works.
Pauline isn't alone. She's part of a cohort proving Phase 2 out right now. The companies that move fast on this connective tissue are building a structural advantage that will help them grow at a rapid pace, long-term.
Insights: The Challenge Ahead
SaaS created a visibility problem slowly. For the better part of a decade, most finance teams couldn't tell you, without hours of manual work, what they were spending on software, until tools like Spendesk came on the market. AI is creating the same conditions, at a different velocity. The structural exposure compounds in months, not years.
SaaS created a visibility problem slowly. For the better part of a decade, most finance teams couldn't tell you, without hours of manual work, what they were spending on software, until tools like Spendesk came on the market. AI is creating the same conditions, at a different velocity. The structural exposure compounds in months, not years.
To track this, we asked our data team to take a relevant split of companies and analyse their subscription spend data. We did this to ensure there weren’t any very large or very small outliers to skew the results.
Across the companies we analysed — 4,500 names, €427 million in qualifying spend over the past twelve months — 57% were paying for two or more AI tools simultaneously. Not at different points in time, as companies evaluate and move on. Simultaneously, in parallel, often with entirely different teams using different tools to accomplish similar work, and with finance knowing about almost none of it.
)
The procurement process that might have caught this was built for a world where software costs tens of thousands of euros and requires a committee to approve. AI subscriptions require a card and a two-minute signup. The infrastructure has not caught up with the velocity of what is being bought.
Especially when a finance lead at a 150-person cybersecurity company put it plainly during a conversation about how they track software renewals: "The payment will fail — and that's usually the first trigger for us to go and change something. We don't get any information around that." That response — recognition, followed immediately by resignation — is now consistent enough across finance leaders that it tells me something broader is happening.
Why This Moment Is Different
SaaS made this problem visible slowly enough that companies could adapt. Finance teams gradually learned to track software spend, build procurement processes, and negotiated contracts.
AI is compressing that timeline.
)
The answer is not to slow adoption. The companies learning to use AI well are building genuine advantages, and they should be moving fast. But speed without visibility is risk.
The answer is to bring the same financial intelligence to AI spend that disciplined finance teams now apply to everything else. To connect the tools to the financial stack rather than letting them proliferate outside it. To know what you have, what it costs, and what it should cost relative to a company like yours.
Connectivity to the financial stack is the first requirement. Knowing what's being spent, in real time. Then knowing what others are paying for the same tools, and having the data to negotiate from.
The crossover in spend has arrived. The intelligence layer is what matters now and it’s what we’re working on next.
Curious how Spendesk works?
Try an interactive demo to see spend control and approvals end-to-end.
Get a free tour
)
)
)
)
)