For years, data teams fought for budget the same way: deck after deck justifying headcount, licenses, and infrastructure costs against business outcomes that were often hard to quantify. That dynamic just changed. According to dbt Labs' 2025 State of Analytics Engineering report, data budgets grew 30% year-over-year—more than three times the prior year's growth rate of 9%. And for the first time, AI tooling ranked as the single highest investment priority across data organizations. This is not a blip. It is a structural shift, and understanding what is driving it matters whether you are a CDO making the case for more resources or a business leader trying to understand why your data team keeps asking for more money.
The Numbers Behind the Surge
The dbt Labs report surveyed thousands of data practitioners across industries and company sizes. The headline figure—30% budget growth—was striking on its own, but the detail beneath it tells a more important story.
Prior years saw data budgets grow modestly, in line with general IT spend, roughly 7-10% annually. The jump to 30% is not a gradual trend accelerating. It is a step change, and survey respondents were clear about the cause: AI initiatives are demanding data infrastructure investments that organizations had been deferring for years.
30%
YoY budget growth in 2025
#1
AI tooling as investment priority
3×
Faster than prior year's growth rate
Source: dbt Labs' 2025 State of Analytics Engineering Report, published April 2025. The report surveyed data practitioners across company sizes, industries, and geographies.
Why AI Is the Forcing Function
Every major AI initiative—whether it is a customer-facing chatbot, an internal copilot, a predictive model, or an autonomous agent—runs on data. Not data in the abstract sense, but clean, governed, well-modeled data that has been transformed, tested, and made accessible in a reliable way. The AI promise is obvious. The prerequisite is less glamorous: your data stack has to be ready for it.
Most data stacks are not. Years of underinvestment have left organizations with fragmented pipelines, inconsistent metric definitions, siloed data sources, and warehouse schemas that made sense when one analyst built them but are now technical debt. When a company tries to build an AI feature on top of that infrastructure, the AI work reveals every gap that existed but was never prioritized.
This is why budget is flowing now. Executives who spent years skeptical of "data for data's sake" are suddenly motivated. They see competitors moving with AI, they understand that AI requires good data, and for the first time they are connecting the two explicitly. The data team's budget request no longer needs to justify itself on reporting efficiency alone—it can be tied directly to the organization's AI roadmap.
Where the Money Is Actually Going
Budget growth does not mean organizations are spending randomly. The report and broader market trends point to four clear areas absorbing the new investment.
AI Tooling and LLM Integration
The most obvious category. Snowflake Cortex, dbt's AI-assisted features, Databricks' Mosaic AI, and a growing ecosystem of AI-native analytics tools all require new licenses, infrastructure, and the engineering time to integrate them. Organizations are buying both platforms and the services to stand them up correctly.
Data Quality and Governance
AI models trained on bad data produce bad outputs—confidently. This has pushed data quality from a background concern to an urgent priority. Spending on observability tools (Monte Carlo, Anomalo, Elementary), data contracts, and governance frameworks has accelerated as organizations recognize that AI amplifies both the value of good data and the damage from bad data.
Transformation Infrastructure (dbt, Pipelines)
Raw data that has not been transformed into clean, modeled tables is not useful to AI systems. Investment in dbt Cloud, Fivetran, Airbyte, and other pipeline tools has grown as organizations expand the coverage of their transformation layer—more sources, more models, more tests, and more of the codebase under version control and CI/CD.
Talent: Analytics Engineers and AI/Data Specialists
Tooling alone does not build a data stack. The demand for analytics engineers—people who sit at the intersection of data engineering and business analysis—has grown faster than supply. The dbt report noted that analytics engineering as a discipline has matured significantly, with more teams formalizing the role and budgeting for it explicitly rather than treating it as a side function of a software engineer or a BI analyst.
The Risk of Waiting
For every organization increasing its data investment, there are others holding back—waiting to see how AI develops, debating whether the ROI is proven, or deprioritizing data infrastructure in favor of AI features themselves. This tradeoff tends to backfire.
AI features built on top of underprepared data infrastructure fail in predictable ways. Natural language querying returns wrong answers because there are no governed metric definitions. Predictive models drift because pipelines are unreliable. Customer-facing AI features get delayed because data access controls were not designed for the new use case. The development teams blame the data teams. The data teams point to years of underinvestment. Both are right.
The compounding debt problem
Unlike software features, data infrastructure debt compounds. Every month that pipelines go untested, metric definitions remain ungoverned, and data models stay undocumented is another month of inconsistency baked into the systems that AI will eventually consume. The cost to clean it up later is always higher than the cost to build it right initially—and the AI use cases that require good data do not wait.
How to Make the Case Internally
If you are a data leader watching this budget shift happen at other organizations but struggling to unlock the same investment internally, the framing has changed in your favor. Here is how I see the most effective arguments landing right now.
Tie every data request to a specific AI initiative
Executives understand AI spending. "We need to improve our data pipeline reliability" is abstract. "Our AI assistant returns incorrect revenue figures because our pipeline has no data quality tests, and that is blocking the product roadmap" is concrete. Find the AI project leadership cares about and connect your infrastructure needs directly to it.
Quantify the cost of data problems, not just the cost of solutions
How many engineering hours per sprint are spent debugging bad data? How many stakeholder questions go unanswered each week because the data team is backlogged? How many decisions are made on stale or inconsistent reports? Putting a number on the status quo makes the investment ask feel proportionate rather than speculative.
Reference the market signal explicitly
The dbt Labs report is public and credible. Showing that peer organizations across your industry are increasing data budgets by 30% reframes the conversation from "our data team wants more resources" to "we may be falling behind a market-wide investment shift." Executives respond to competitive framing.
What This Means for the Next 12 Months
The budget surge reflects a recognition that has been a long time coming: AI is not a separate capability that companies can bolt onto their existing systems. It runs on data, and the quality of that data determines the quality of the AI. Organizations that have built rigorous data infrastructure—clean pipelines, governed metrics, well-modeled warehouses—are positioned to move fast on AI. Those that have not are discovering that the first step in their AI roadmap is actually a data infrastructure project.
The dbt Labs and Fivetran merger announced in 2025 is one signal of how the market is consolidating around this reality: end-to-end data infrastructure, from ingestion to transformation to semantic layer, is becoming a single category rather than a collection of point solutions. Snowflake and Anthropic's $200 million expanded partnership is another: the assumption is that AI models will be querying governed Snowflake data, and the value of that integration depends entirely on the data being trustworthy.
For data teams, this is a moment of genuine leverage. The investment case that struggled to compete with product development and marketing for budget is now tied directly to an initiative every executive in the company is prioritizing. The work has not changed—building reliable, well-governed data infrastructure has always been the goal. What has changed is that the rest of the organization now understands why it matters.
The bottom line
A 30% budget surge is not a temporary wave driven by AI hype. It is the market correcting years of underinvestment in data infrastructure at exactly the moment when that infrastructure has become a competitive prerequisite. Whether you are accelerating investment or still building the case internally, the direction is clear.
Is your data stack ready for AI?
I help data teams audit their current infrastructure and build the roadmap to close the gaps—before an AI initiative reveals them the hard way.
Book a Free Data Audit