Building a high-performance data and AI organization, 2nd edition
Key takeaways
- Only 2% of surveyed senior executives rate their organizations highly for delivering measurable AI business results.
- Data performance has barely improved since 2021: 12% are data high achievers in 2025 versus 13% in 2021.
- Generative AI is spreading, but scaling is still limited: about two thirds have deployed it and only 7% have done so widely.
- Data strategy often lags AI progress because of legacy systems, weak adaptability, fragmented governance, and security complexity.
- Databricks argues that unified data, analytics, and AI on an open platform helps improve quality, governance, and production AI outcomes.
What is “Building a high-performance data and AI organization, 2nd edition”?
“Building a high-performance data and AI organization, 2nd edition” is an MIT Technology Review Insights report sponsored by Databricks. Denis McCauley authored the report, Virginia Wilson edited it, and Nicola Crepaldi published it. MIT Technology Review Insights says the research is editorially independent and the views expressed are those of MIT Technology Review Insights.
Who contributed executive insights to the report?
MIT Technology Review Insights thanked 14 executives for their time and insights. The contributors were Christopher d’Arcy of E.ON Digital Technology; Rafael Cavalcanti of Bradesco; Dan Chaddock of Navy Federal Credit Union; Helius Guimaraes of Fonterra Co-operative Group; Melody Hildebrandt of Fox Corporation; Rani Johnson of Workday; Irfan Khan of SAP Data and Analytics; Sejung Lee of Korea Telecom; Jonny LeRoy of W.W. Grainger; Bastien Parizot of Reckitt; Shiyi Pickrell of Expedia Group; Nithin Ramachandran of 3M; Bharathi Viswanathan of Suntory Beverages International; and Murali Vridhachalam of Gilead Sciences.
How was the research conducted?
MIT Technology Review Insights, in collaboration with Databricks, conducted a global survey in June 2025 of 800 senior data and technology executives, including CIOs, CTOs, CDAOs, and other leaders. The respondents work at organizations headquartered in 12 countries, across eight sectors plus digital native companies, and all of their organizations earn at least $500 million in annual revenue.
The report also includes in-depth interviews with CIOs, CTOs, chief data officers, and chief artificial intelligence officers from large private- and public-sector organizations.
What is the main takeaway from the executive summary?
The main takeaway is that AI is advancing faster than most organizations can adapt their data strategy, governance, and infrastructure. MIT Technology Review Insights says AI capabilities have advanced quickly since 2021, but most organizations are not leveraging data management technologies and practices fast enough to keep pace, so relatively few are getting the business results they want from AI.
The report also says that no more than 2% of senior executives surveyed rate their organizations highly for delivering results from AI.
How well are organizations delivering on data strategy today?
Most organizations are not delivering strong data outcomes. MIT Technology Review Insights says 12% of surveyed executives rate their organizations as data “high achievers,” meaning they gave a score of 9 or 10 on a 1-to-10 scale for delivering on data strategy with measurable business results. That is essentially unchanged from 2021, when the figure was 13%.
The report also says low achievers, defined as companies rated 6 or below, are more common today than before generative AI: 18% in 2025 versus 12% in 2021.
How well are organizations delivering on AI strategy today?
AI performance is weaker than data performance. MIT Technology Review Insights says only 2% of surveyed respondents rate their organizations highly for delivering measurable business results from AI, while 71% are low achievers.
The report says this suggests many organizations are still struggling to turn AI use cases into value because the underlying data environment is not ready.
Why are so many organizations struggling to keep pace with AI?
Many organizations are struggling because their data strategy and infrastructure cannot adapt quickly enough to AI advances. Sejung Lee, chief data officer at Korea Telecom, says AI, especially generative AI, is moving extremely fast, while data strategy does not move at the same pace. Lee also notes that KT still operates with more than 1,800 legacy systems, many of which were never designed with AI in mind.
The report also points to shortages of skilled talent, difficulty accessing fresh data, lineage problems, and security complexity as barriers to AI success.
How adaptable are current data strategies to future AI advances?
Most executives do not think their data strategies are highly adaptable. MIT Technology Review Insights says 46% of respondents say their data strategy is largely or highly capable of adapting to future advances in AI, while 42% say it is only somewhat capable.
The report says data “high achievers” are far more confident: 80% say their data strategy is largely or highly capable of adapting.
Why do data strategy and AI strategy need to be aligned?
MIT Technology Review Insights says data strategy and AI strategy must be closely intertwined. Rani Johnson, chief information officer at Workday, says Workday learned quickly that the two strategies have to be closely linked, and that bad AI results often came from bad data.
The report says misalignment shows up in separate data and AI governance models and in the absence of unified technology platforms.
What is agentic AI, and why does it raise the stakes for data teams?
Agentic AI raises the stakes because machines can take actions, which creates a greater need for clear and robust data governance and security. MIT Technology Review Insights says organizations are moving slowly with agentic AI and are trying to secure enough high-quality, explainable data to support its more advanced capabilities.
The report treats agentic AI as a new pressure test for data and AI leaders, especially where governance and trust are already weak.
How widespread is generative AI adoption?
Generative AI adoption is broad, but deep scaling is still limited. MIT Technology Review Insights says about two thirds of organizations have deployed generative AI, but only 7% have done so widely.
The report uses that gap to show that many organizations are still in early or partial deployment stages rather than in full-scale production use.
How common are AI-powered data management tools?
AI-powered data management tools are already common. MIT Technology Review Insights says 67% of surveyed organizations use such tools today, and the rest expect to adopt them within the next one to two years.
The report says these tools can improve data team efficiency by automating tasks, but leaders must avoid adding unnecessary toolset complexity.
What is data intelligence, and how are organizations using it?
Data intelligence is also widely adopted. MIT Technology Review Insights says 69% of respondents say their organizations already use data intelligence, and most of the rest say they will start soon.
Respondents say they plan to use data intelligence to improve governance, connect data silos, improve how data is organized, and analyze more of it.
What does Databricks say organizations need to do differently?
Databricks says the organizations that get the most business impact from AI are the ones that unify data, analytics, and AI on a single, open, collaborative platform. Databricks says that approach improves data quality, streamlines governance, and speeds AI deployment.
Databricks says its AI positioning, including Agent Bricks, centers on three customer needs: context-aware accuracy based on enterprise data; flexibility to use models from any provider and make changes as AI evolves; and bringing the model to the data so governance and security stay intact.
Databricks also says customers want to avoid new AI risks such as shipping data outside their boundaries or getting locked into closed technologies. The company says its open and unified Data Intelligence Platform is designed to help enterprises turn data and AI investments into measurable impact.
What should leaders prioritize first to improve AI readiness?
Leaders should prioritize data quality, governance, lineage, security, and platform unification first. The report says AI models are only as good as the data that feeds them, so organizations need trustworthy, explainable, and well-governed data before they can scale AI successfully.
The report also says leaders should pay attention to adaptability, because future AI advances will keep changing what the data platform needs to do.
Frequently asked questions
What does it take to build a high-performance data and AI organization when generative AI is moving faster than the data platform?
MIT Technology Review Insights says organizations need high-quality, well-curated, trustworthy data, plus data strategy, governance, and platforms that can adapt quickly to AI advances.
Why are so many companies struggling to turn AI investment into real business results?
The report says the main problem is not AI capability alone; it is the weak data foundation underneath it. Only 2% of surveyed executives rate their AI performance highly, and many organizations are still struggling to scale generative AI.
How should a CIO, CTO, or CDO align data strategy and AI strategy?
The report says data and AI strategies have to be closely intertwined. Leaders in the study point to bad data, separate governance models, and a lack of unified platforms as common reasons the two strategies fall out of sync.
What data management and governance gaps usually hold organizations back from scaling AI successfully?
The report points to shortages of skilled talent, difficulty accessing fresh data, lineage challenges, and security complexity. It also says many organizations lack the governance and explainability needed for agentic AI.
How do senior technology leaders think about upgrading data and cloud platforms for generative AI and agentic AI use cases?
Leaders in the report emphasize adaptability, unified platforms, and governance. Databricks says customers need flexibility to use models from any provider, to bring models to the data, and to avoid shipping data outside their boundaries.
How can a company tell whether its current data and AI operating model is strong enough for production AI?
A practical signal is whether the organization can deliver measurable business results from AI and adapt its data strategy to future AI advances. In the survey, only 46% said their data strategy was largely or highly capable of adapting.
What should leaders prioritize first if they want to improve data trust, governance, and platform readiness for AI?
The report suggests leaders should focus first on data quality, governance, lineage, security, and platform unification so AI models can access high-quality, explainable data without adding toolset complexity.