Fivetran today released The AI tax: The hidden cost when AI isn’t connected to your data, a new report that found nearly 4 in 10 companies (39%) spend at least a quarter of their AI budget addressing data problems, including fixing data quality issues, rebuilding integrations and reprocessing data. The findings reveal how this hidden cost, termed the “AI tax,” is diverting technical resources, delaying AI initiatives and limiting returns on AI investments. The research is based on a survey of 500 senior IT, data and AI leaders across the United States, Europe and Asia-Pacific.
“Companies are spending millions on AI, but too much of that investment is going toward fixing the same data problems over and over again,” said George Fraser, CEO and Co-founder of Fivetran. “AI is becoming the primary consumer of enterprise data, and the job of the data team is changing with it. Instead of building dashboards and pipelines for every new request, data teams need to connect AI to trusted enterprise data and give it the business context to understand that information. That’s how companies can stop rebuilding the same infrastructure and start getting more value from their AI investments.”
The hidden costs of AI are adding up
As businesses deploy AI agents across more workflows, demand for trusted, accessible enterprise data is increasing. Many organizations, however, are investing heavily in models and applications without the data infrastructure needed to support them. The study found that 83% of organizations invested at least $1 million in AI initiatives over the past year, with more than half (54%) spending $5 million or more. Despite that investment, only 37% report having multiple AI initiatives operating at scale.
Much of the burden comes from the work required to make enterprise data usable for AI. Two-thirds of organizations (66%) identify data quality or integration challenges as leading sources of unexpected work, delays, or costs. More than half (58%) report that technical teams spend at least a quarter of their time supporting AI on avoidable data work, including remediation, reprocessing, retraining models, and rebuilding integrations.
U.S. companies face the highest AI tax
U.S. organizations report the highest AI tax among the regions surveyed. Nearly half (45%) say at least a quarter of their AI spending goes toward avoidable work, compared with 34% in Europe, the Middle East and Africa (EMEA) and 37% in Asia-Pacific (APAC). For example, a company spending $5 million annually on AI would lose $1.25 million if a quarter of that budget went to avoidable work.
The burden significantly impacts technical resources. More than 1 in 5 U.S. organizations (22%) report that technical teams spend at least half their time supporting AI on avoidable data work, compared with 14% in EMEA and 11% in APAC.
U.S. organizations also see the greatest potential for improvement. Half (50%) estimate that eliminating avoidable data work could accelerate AI time to value by more than 25%, compared with 35% in EMEA and 38% in APAC.
Custom integrations linked to higher AI costs and longer delays
Organizations relying primarily on custom integrations report greater AI spending losses than those using standardized platforms. More than half (53%) lose at least a quarter of their AI spending to avoidable work, compared with 40% of organizations using standardized integration platforms.
The same organizations are also twice as likely to experience data-related project delays lasting more than a month (40% vs. 20%). Across all respondents, two-thirds (66%) report needing moderately or significantly more data work than originally planned after AI projects began.
Connecting AI to trusted data accelerates returns
The research suggests that reducing avoidable data work could help companies realize value from AI investments faster. Nine in 10 organizations estimate that eliminating these inefficiencies could improve AI time to value by more than 10%, while 41% anticipate improvements greater than 25%.
Companies spending less on avoidable data work also report stronger returns on their AI investments. Nearly two-thirds (62%) of organizations losing less than 10% of their AI budgets to avoidable work say their AI returns exceeded expectations, compared with 40% of those losing half or more.
Data infrastructure is also emerging as a leading investment priority. Asked where they would invest additional AI budget over the next 12 months, nearly one-third (31%) of respondents chose their data foundation, ahead of talent (17%) and AI models (16%).
The full report, The AI tax: The hidden cost when AI isn’t connected to your data, is available here.
Research Methodology
The study is based on a global survey of 500 senior IT, data, and AI/ML leaders conducted between August 24 and September 7, 2026. Respondents represented organizations across the United States, Europe, and Asia-Pacific, spanning manufacturing, financial services, technology, retail, healthcare, and other industries.
The research examined AI investment and deployment, data foundations and integration, avoidable work and delays, and organizational performance and ROI. The AI tax is defined as the self-reported share of AI/ML spending consumed by avoidable work over the preceding 12 months.
About Fivetran
Fivetran is the company that connects your AI to your data, so customers can act with confidence. Fivetran moves data from more than 750 sources and gives it the definitions, logic and lineage that make it trustworthy. It connects that data to any agent, tool or team through open infrastructure, giving customers the freedom to choose their clouds, data platforms and AI tools while maintaining control of their data. Over 13,000 global customers, including Anthropic, Canva, OpenAI, Perplexity, Siemens and Verizon, rely on Fivetran, while more than 100,000 data teams use dbt. Fivetran’s offerings include Fivetran data movement, dbt, Managed Data Lake and dbt Wizard, with dbt Core and its open-source community remaining central to the company’s commitment to open infrastructure.
Learn more at Fivetran.com and getdbt.com, or follow Fivetran on LinkedIn and X, and both Fivetran and dbt on YouTube. We’re just getting started.
View source version on businesswire.com: https://www.businesswire.com/news/home/20261112409708/en/
Media gallery
