Why companies keep pouring money into failing AI projects

A recent survey suggests that most companies are frustrated with their lack of progress in generative artificial intelligence, yet they keep throwing more and more money at the technology.

Is this money drain going to continue in 2025, or are we turning the corner and correcting some of the major problems? We're going to talk about that and more on this episode of Today in Tech. Hi, everybody.

Welcome to Today in Tech. I’m Keith Shaw. Joining me on the show today is Brett Roscoe—Senior Vice President and GM of Cloud Data Governance and Cloud Ops at Informatica—and Blake Andrews, Chief Data and Analytics Officer at Independent Financial. Welcome to the show, gentlemen.

I want to talk about a lot of these issues. We’ve got a lot of data from the 2025 CDO Insights report, which surveyed 600 data leaders at companies with more than $500 million in revenue across the U.S., U.K., Europe, and Asia Pacific.

What was the main goal of the survey?

Brett: This year, we really focused on: What are the stumbling blocks? What are your investment plans? What are the issues or concerns you have around your AI and GenAI projects?

Keith: It certainly revealed some insightful information about progress being made and the scale of investments. But clearly, customers are still struggling to move forward. Blake, you're involved in this—not as an author of the report, but in what capacity?

Blake: I’m a member of the Informatica CDO Executive Advisory Board, providing industry insights and validating some of the report’s findings. I'm representing the IT and data side—essentially the CDOs of the world—and helping to confirm these answers.

So yeah, I've worked with a lot of large financial enterprises trying to move their GenAI projects forward. I’ve talked with Informatica before, and I try to be a strong customer voice. Keith: That’s great. When you both look at this year’s survey compared to 2023, were there any surprises?

Or do the results validate the assumptions you already had? Brett: A couple of things continued to validate—like the fact that big investments are still happening. About 87% of respondents said they plan to increase their GenAI investments from 2024 to 2025, just like they did from 2023 to 2024.

The biggest "aha" moment for us came in two areas. First, many organizations are struggling to move their GenAI projects beyond proof of concept or pilot phases into full production. Second, 97% said they’re having trouble proving ROI—justifying these projects to their business units and leadership.

Despite seeing value and continuing to invest, demonstrating that value remains a major hurdle.

Keith: Blake, when you saw the results, did they validate your experience at Independent Financial? Or did you feel like an outlier? Blake: Some findings definitely resonated with me, and with the conversations I’ve had with other CDOs through various networks.

When you start aggregating those anecdotal perspectives, they really come to fruition in the results of this survey.

Keith: Earlier, Brett, you mentioned that about two-thirds of respondents said they haven’t been able to transition even half of their GenAI pilots to production. We've heard a similar 30% number from Gartner. What are the biggest reasons for this?

Brett: We had a fairly even distribution across five core reasons, which were very consistent across respondents. Data quality – Your GenAI outputs are only as good as the data going in. Privacy and security – Protecting personal, sensitive, or regulated information.

Skills and company culture – There's a need for technical skills and a supportive culture to adopt GenAI. Regulatory compliance – Adapting to evolving rules, especially with new EU and state-level regulations. Responsible AI use – Ensuring ethical, unbiased, and safe use of the technology.

Keith: Would those five challenges be evenly represented on a bar chart, or do some stand out? Brett: It’s pretty even. Around 43% cited security as an issue, and 43–44% pointed to data quality. Privacy and quality were the top concerns, but the rest weren’t far behind.

Keith: Do companies feel they can tackle all five challenges, or are they prioritizing security and privacy first?

Blake: Quality will always be something you have to address—it’s fundamental to data governance. AI acts like a magnifying glass, amplifying any data quality issues you already have. So companies need to be very tuned into those issues. Another key area is risk and compliance.

Companies are learning that they need to bring privacy, legal, and risk experts into the process early—not at the 11th hour. When these stakeholders are brought in too late, they don’t understand how the project evolved, and that creates visibility challenges.

Keith: Yeah, bringing those teams in at the end seems like a recipe for disaster. And it’s not something you solve once—it’s project-by-project, right? Blake: Exactly.

There’s administrative overhead, and it’s even more strained when companies run multiple pilots at once. Some projects are external-facing and pose a higher risk, so they get more attention. Internal ones may get less scrutiny but still need governance.

Another issue identified was lack of maturity and interoperability in AI tech, which stood out in this survey. Keith: That’s new. We often hear about security, compliance, and ROI—but maturity and interoperability are interesting. Are companies waiting for better tools, or struggling to make many tools work together?

Brett: The space is evolving so fast—it’s hard for teams to keep up. Some companies believe they need 10 different tools to manage data and AI projects, which creates compatibility issues. That’s why we at Informatica push for a platform approach with strong interoperability.

But even then, integrating with outside AI tools still requires technical skill. Companies are trying to figure out how to choose the right partners and tools to create a cohesive, streamlined system. Blake: Right, and that ties directly into ROI.

The more fragmented your toolset, the harder it is to prove return on investment. Keith: Are finance teams pushing back, or is the difficulty more about proving value?

Blake: It's a mix. Expectations around ROI are extremely high. Pop culture has trained us to think AI is this hyper-capable tool—but in reality, we’re still in the growing pains phase.

Costs skyrocket when you move from pilot to enterprise scale, and some companies rushed into projects without well-defined use cases. They’re now trying to retrofit solutions to problems, which rarely works well.

Keith: So it’s like, “We have all this data—let’s throw AI at it and see what happens.” Blake: Exactly.

There’s also that fear of missing out. Companies don’t want to fall behind, especially if competitors are innovating with AI. The pressure to act quickly is real. Keith: But it can’t just be about throwing money at the problem, right? Brett: Right.

You need holistic investment. Our survey found 86% are investing in mature data management capabilities. GenAI demands robust, governed data sharing practices. So companies are scaling not just their AI teams, but their governance and management frameworks, too. It’s about generating safe, ethical, and valuable outcomes.

Keith: Will we see a pullback eventually? Brett: Probably.

Investments can’t go up forever. But I don’t think we’re there yet—we’re still in the ramp-up phase. Keith: Blake, do companies also fear failure—not just wasted spending, but reputational damage from a flawed AI rollout? Blake: Yes, definitely.

There's a difference between failing fast in a pilot and launching something risky to the public. Agile methodologies are helpful here—fail early, learn quickly. The biggest failures happen when you skip governance and risk assessment. Bringing compliance and legal teams into the process from the start is crucial.

You have to think about how customers interact with AI safely and responsibly. Keith: So the idea is: better to fail in a pilot than in public. That makes sense. Brett: Exactly.

The survey shows that most customers are testing with POCs and pilots before moving to production, which is encouraging. You want to discover flaws early, before pouring money into a project. Keith: You mentioned earlier the pressure from the C-suite.

Do data leaders feel like they’re the “stick in the mud,” trying to pump the brakes?

Brett: There’s definitely a push-pull dynamic. The C-suite is eager to go fast, while legal and compliance teams want to move cautiously. Companies need governance committees and strong communication to strike that balance.

Blake: Yes, managing expectations is key. Mature programs can flex the frameworks they’ve already built. Others are scrambling to catch up. The good news is that the C-suite is finally pulling for governance and data foundations, instead of needing to be pushed.

Keith: Is there time for disillusionment, or is the pace too fast to slow down? Blake: You nailed it. The hype cycle is real—but before disillusionment sets in, a new breakthrough like agentic AI or GPT hits the scene and reignites the excitement. Brett: Totally agree.

Even governance can now be accelerated with AI—using it to automate policy creation, data classification, and more. We talk about AI for data management and data management for AI. Both sides of the equation are evolving quickly. Keith: That keeps you all very employed!

And as you said, data’s not going away. Blake: Exactly. It’s just one hype cycle layered on top of another. The next big thing is always just around the corner. Keith: That’s a great point to end on. The optimism is still there.

Thanks again, guys, for joining me on the show. That’s all the time we have for this week’s episode. Be sure to like the video, subscribe to the channel, and drop your thoughts in the comments. Join us every week for new episodes of Today in Tech.

I’m Keith Shaw—thanks for watching!