Nearly four in ten companies spend at least a quarter of their AI budget on resolving data issues. Custom integrations drive costs even higher.
This is according to Fivetran research. The company refers to this as the “AI tax”: the hidden costs that arise when AI isn’t properly connected to business data. According to the study, 39 percent of companies spend at least a quarter of their AI budget on fixing data quality, rebuilding integrations, and reprocessing data.
The budgets are substantial. Of the organizations surveyed, 83 percent invested at least $1 million in AI last year, and 54 percent invested $5 million or more. Yet only 37 percent have multiple AI initiatives running at scale.
Technical teams also spend time on avoidable work. At 58 percent, these teams spend at least a quarter of their AI time on remediation, retraining models, and rebuilding integrations. Two-thirds cite data quality or integration as the main source of unexpected work, delays, or costs.
“Companies are spending millions on AI, but too much of that investment is going toward fixing the same data problems over and over again,” says George Fraser, CEO of Fivetran. According to him, data teams need to connect AI to reliable business data and provide the right business context, rather than building dashboards and pipelines for every new query.
The cost is highest in the U.S., where 45 percent say at least a quarter of AI spending goes toward avoidable work, compared to 34 percent in EMEA and 37 percent in APAC. At 22 percent of U.S. organizations, as much as half of team time is spent on this type of work (EMEA: 14 percent, APAC: 11 percent).
Custom solutions cost more
What makes the difference? According to Fivetran, it’s primarily the integration approach. Among organizations that rely heavily on custom integrations, 53 percent lose at least a quarter of their AI budget, compared to 40 percent among users of standardized platforms. They are also twice as likely to experience data-related delays of more than a month (40 percent versus 20 percent).
Those who lose less also see a higher return on investment. Among organizations that lose less than 10 percent to avoidable work, 62 percent say AI exceeded expectations. Among those losing half or more, that figure drops to 40 percent. When asked about additional AI budgets for the next twelve months, 31 percent prioritize data infrastructure, ahead of talent (17 percent) and AI models (16 percent).