Companies in eight countries spent a combined $63.4 billion more on technology infrastructure in 2025 than they had budgeted, according to a study commissioned by cloud storage provider Wasabi Technologies and released November 10. German companies alone overspent by €7.1 billion. The digest versions for other markets put France at €4.7 billion and Italy at €2.8 billion. The study also covered the US, UK, Japan, Australia and New Zealand.
How the numbers were built
Sapio Research surveyed 690 senior decision makers in July and August 2026, 100 of them in Germany. The Centre for Economics and Business Research (CEBR) then scaled the survey answers into national estimates and modeled what the money could have funded. For Germany, CEBR says the unplanned spend could have supported €2.5 billion in productive investment, 10,878 jobs and €2.4 billion in long-term economic output.
Those second-order figures are model outputs, not observed losses. A sample of 100 respondents per country is small for national extrapolation, and the estimates should be read as an order of magnitude.
AI is the hardest cost to forecast
One in five respondents describes the actual infrastructure costs of AI projects as unpredictable. Among companies hit by unexpectedly high infrastructure costs, 75% report added cost pressure, and many respond by delaying hires, training or new technology projects. Companies that try to switch providers or add capacity also report hurdles in doing so.
Wasabi’s president and COO Marty Falaro framed it as companies having ambition and being caught out by costs they could not foresee.
Why Wasabi commissioned it
Wasabi sells flat-rate object storage without egress or API request fees, so a study pinning overruns on unpredictable usage-based charges lines up with its pitch against the large cloud providers. The survey questions and the full country breakdown were not published with the release. Still, the direction matches what the hyperscalers’ own capital spending shows: AI workloads move a lot of data, and data movement is where cloud bills tend to surprise. Our AI capex tracker follows the supply side of that spending.
