State of Data Teams 2026: AI Adoption in Data

· Hex

2 min read Original article ↗

About this survey: In December 2025, Hex re-surveyed 65+ data leaders — VPs of Data, directors, analytics engineers, and individual contributors — about their priorities, investments, and blockers. Respondents span company sizes from 80 to 1,000+ employees.

Just six months after our first State of Data Teams Report , the shift was clear: attitudes toward AI moved fast.

In this most recent survey, “AI and automation” jumped to a top goal for data teams. Last Spring, only 3 respondents mentioned AI as a goal for their team.

But it’s not just a goal — data teams are already putting hours into activating AI. In a list of focus areas, AI ranked second, with the number of respondants calling it an important area increasing by 24% compared to the last study.

But moving full speed ahead on AI doesn't mean the road is entirely smooth; the majority of respondents said that data trust is the #1 concern around AI adoption (which was cited nearly twice as much as any other concern). To address trust and accuracy, data teams reported investing in a variety of governance layers for AI to be more accurate. Semantic layers, once controversial, are seen as essential to this shift.

And despite concerns about AI replacing data work, the opposite is happening. 58% of respondents reported their data team grew, up from 52% in spring — a signal that AI is reshaping data work, not eliminating the need for it.

Let's get into the details.

Key findings

  • AI as a data team priority surged 575% in six months: only 4% of leaders cited it as a top goal in Spring 2025; 27% did by December 2025

  • 31% of data leaders cite trust and data accuracy as their #1 AI concern — nearly 2× the next most common barrier

  • 58% of data teams grew by late 2025, up from 52% in spring; only 4% reduced headcount (down from 9%)

  • AI tooling jumped from 6th to 2nd in data team focus areas, a 24% increase in six months

  • 0% of data leaders predict they'll avoid AI entirely

  • Semantic layers are now consensus: no respondent questioned their value in fall (vs. a minority of doubters in spring)