The chief data officer (CDO) has evolved from a niche compliance role to one of the most important positions for AI adoption. These executives are now at the intersection of data governance, AI strategy, and workforce readiness. Their decisions will determine whether companies move from AI pilots to production scale or remain in experimental mode.
That’s why Informatica’s third annual survey – CDO’s largest-ever survey of 600 executives around the world, specifically on AI readiness – carries special weight. The findings reveal a dangerous disconnect that explains why many organizations struggle to scale AI beyond pilots. While 69% of companies have deployed generative AI and 47% are running agent AI systems, 76% admit that their governance frameworks have not kept up with how employees are actually using these technologies.
This study reveals what Informatica calls the “paradox of trust” and explains why data leaders are dangerously overconfident in their ability to respond to AI. Organizations deployed generative AI systems faster than they could build the governance and training infrastructure to support them. As a result, while employees generally trust the data that powers AI systems, organizations recognize that employees lack the literacy to question that data and use AI responsibly. 75% of data leaders say they need to upskill their employees in data literacy. 74% need AI literacy training for daily work.
“The only gap now is can we trust the data that will unleash our agents,” Informatica CIO Graeme Thompson told VentureBeat. “Agents will do what they’re supposed to do if they’re given the right information. I think there’s a gap there because there’s such a lack of trust in the data.”
Why infrastructure isn’t the bottleneck for data and AI
GenAI adoption has increased from 48% a year ago to 69% today. Almost half (47%) of organizations are now running agent AI (systems that autonomously perform actions in addition to generating content). This rapid expansion has created a race to acquire vector databases, upgrade data pipelines, and expand computing infrastructure.
But Thompson denies that the infrastructure gap is the main problem. The technology exists and it works. This limitation is organizational, not technical.
“The technology, the infrastructure, that we have available to us right now is so much more than that, and that’s still not the issue,” Thompson said. He likened the situation to amateur athletes blaming their equipment. “We have a long way to go before equipment becomes an issue in the room. People chase equipment the same way golfers do. Those golfers are obsessed with new drivers and new putters that cure their lack of physical ability to hit a golf ball straight.”
Survey data supports this. When asked about their investment priorities for 2026, the top three were all about people and process. Data privacy and security (43%), AI governance (41%), and employee upskilling (39%).
Five hard lessons for corporate CDOs
The combination of survey data and Thompson’s implementation experience reveals concrete lessons for data leaders looking to move from pilot to production.
Stop chasing infrastructure, let’s solve people’s problems
The trust paradox exists because organizations can deploy AI technology faster than they can train people to use it responsibly. 75% need to upskill in data literacy. 74% require AI literacy training. The technology gap is a talent gap.
“It’s much easier to get employees who know the company, the data, and the processes to learn AI than to bring in an AI person who knows nothing about the company and teach them about the company,” Thompson says. “And just like data scientists are very expensive, AI talent is also very expensive.”
Make the CDO an executive function rather than an ivory tower
Thompson builds Informatica so that the CDO reports directly to him as the CIO. This makes data governance an execution function rather than a separate strategic layer.
“This is a deliberate decision based on the fact that the function is not an ivory tower function, but a function that gets things done,” Thompson said. This structure allows data teams and application owners to share common priorities through a common manager. “If you have a common boss, their priorities should align. If they don’t, it’s because the boss isn’t doing his job, not because the two departments aren’t working from the same priority list.”
If 76% of organizations fail to effectively manage the use of AI, reporting structures may be part of the problem. Siled data and IT functions create conditions for pilots that cannot scale.
Increase literacy outside of your IT team
The breakthrough insight is that AI literacy programs need to extend beyond technology teams to business functions. At Informatica, the chief marketing officer is one of Thompson’s strongest AI partners.
“You need that literacy, not just on the technology team, but on the entire business team,” Thompson says.
He said the marketing operations team understands technology and data. You know the answer to the question, “How can I get more value from my limited annual marketing program budget?” That’s not done by adding headcount or increasing Google’s ad spend, but by automating and adding AI to the way that job gets done.
Literacy on the business side creates impetus, rather than driving, AI adoption. Marketing, sales, and operations teams are beginning to demand AI capabilities as they see strategic value in addition to increased efficiency.
Propose AI as a strategic enhancement rather than a cost reduction
Data leaders have spent decades fighting the perception that IT is just a cost center. AI offers an opportunity to change that narrative, but only if CDOs reframe their value proposition from productivity savings.
“I’m very disappointed that even though the capabilities of this new technology were taking off, both as IT and data people, we quickly pivoted and started talking about productivity savings,” Thompson said. “What a wasted opportunity.”
Tactical change: Rather than reducing existing employee numbers, propose AI capabilities that remove employee headcount constraints entirely. This reframes AI from an operational efficiency to a strategic capability. Organizations can expand their market reach, enter new geographies, and test initiatives that were previously cost-prohibitive.
“It’s not about saving money,” Thompson said. “If that’s primarily your approach, your company won’t win.”
First move vertically and scale the pattern
Don’t wait for a full horizontal data governance layer before delivering production value. Choose one high-value use case. Build a complete governance, data quality, and literacy stack for your specific workflow. Validate the results. Then duplicate the pattern to adjacent use cases.
This enables production value to be realized while incrementally building organizational capabilities.
“I think this field is moving very quickly. If you try to solve 100% of the governance problems before you get to the semantic layer problems, before you get to the lexicon problems, you’re never going to get results and people are going to lose patience,” Thompson said.
