AI Treasury · AI Budgets
Unpredictable Tech Bills Reportedly Cost British Firms £3.8 Billion a Year. AI Is the Hardest to Forecast.
AI infrastructure accounts for £1.2 billion, and 36% of firms under budget pressure delayed or cut AI work.
· 2 min read
Wasabi Technologies, a cloud storage vendor, commissioned the Centre for Economics and Business Research (CEBR) to size tech charges British firms could not forecast: £3.8 billion a year, according to a published report.
AI is the biggest single source of that bill and the one businesses say they can least predict. A forecast that carries last year’s AI line forward carries its error forward too.
What makes AI the hardest line to forecast?
About a third of the UK total, £1.2 billion, traces to AI and machine-learning infrastructure, more than any other category, the report said. Across all eight economies in the study, CEBR found that 22% of businesses call their AI infrastructure costs unpredictable. No other layer of the technology stack tops 12%, and the lowest is 8%.
Three drivers lead the AI overspend among British firms that had one, the report said. Unexpected platform and model licensing fees top the list at 66%, ahead of integration, deployment and model operations at 43% and higher-than-expected token, inference or usage-based charges at 36%.
What does the overspend do to AI work?
Nearly three-quarters of British firms, 73%, reported moderate or significant budget pressure from unpredictable costs, the report said. Of those firms, 36% put AI, automation or data science projects on hold or cut them back, a different measure from the usage-charge share above.
Owen Good, CEBR’s head of economics advisory, drew the same line. “The technology generating the most unplanned costs is also among the investments businesses are cutting to pay for them,” he said.
How far does the finding reach?
The UK is one of eight economies in the study, where the overspend comes to $63.4 billion a year. AI and machine-learning infrastructure makes up $23.4 billion, or 37%, of that total, Wasabi said. The survey behind the study reached 690 senior business and IT decision-makers across all eight markets in July and August.
Marty Falaro, Wasabi’s president and chief operating officer, said businesses “need to know what it will cost to store, access and move their data before they commit to an AI project, not after the bill arrives.”
The test for a company’s own plan is which of the three AI drivers it has forecast with a range, and which it carries as a single number.
