IT leaders are discovering that the path to scaling artificial intelligence begins with a strong foundation in data architecture, ensuring that enterprises can trust and leverage their information at scale.

The push to modernize has brought new challenges, from fragmented data silos to inconsistent governance practices. Organizations are realizing that the differentiator will not be access to generic AI models but rather the ability to harness their own structured and unstructured data effectively. This shift is reshaping enterprise strategies and has put the spotlight squarely on the role of architecture as a driver of transformation, according to Marcela Vairo (pictured), vice president of data and AI, Americas, at IBM Corp.

IBM’s Marcela Vairo talks about using data architecture to scale AI.

“We are helping customers to be successful as AI-driven companies,” Vairo said. “What does it mean? It means preparing their data, getting their data architecture-ready, because there is no AI without data. This hasn’t changed. We are helping them to select the best use cases and to think about the return on investment and how to grow and scale AI to transform the way they operate.”

Vairo spoke with theCUBE’s Dave Vellante at Cloudera EVOLVE25, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how strong data architecture enables enterprises to scale AI effectively across hybrid and multicloud environments. (* Disclosure below.)

Why data architecture is key to scaling AI

As enterprises move beyond proofs of concept into large-scale deployment, they are running into new roadblocks tied directly to the state of their data. Without trust in data quality and governance, scaling AI becomes nearly impossible. This growing awareness is changing the way business and technology leaders view their operations and investments, Vairo noted.

“As it evolves and as they start putting AI really into production, AI to work, they’re realizing that if they don’t have the correct data, if they don’t trust their data, if they don’t have governance, they won’t be able to scale,” she said. “I think this concept of data and AI coming up together, it’s clear not only to the IT guys; I think this is a major change. But the line of business is also realizing that they need to figure out that data, they need to trust their data in able to scale with AI.”

Hybrid and multicloud environments are now the norm, further complicating the data landscape. Workloads remain spread across on-premises systems and cloud platforms, requiring a flexible approach that can unify data without disrupting business processes. Companies are recognizing that the true intelligence of their AI models will come from their own unique data, not just from pre-trained external systems, Vairo explained.

“To grow with AI, you have to figure out how to use your data because the differentiator would come by using your own data,” she added. “The data that it’s in your company and outside of that structure and unstructured data, that is where the intelligence for your AI applications and for your AI agents is coming from.”

The untapped potential is staggering, Vairo pointed out. Despite the data explosion, only a small fraction of enterprise information is being utilized for AI today. Most of this lies in unstructured formats, which are harder to access but contain the most opportunity for differentiation and competitive advantage. Unlocking this data requires both governance frameworks and the right architecture.

“More than 90% of the data is unstructured and it’s growing three times faster than structured data,” Vairo said. “It’s really the key. I would say the secret would be how do I leverage all this unstructured data in hybrid environments to fuel my AI agents and my AI applications.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of Cloudera EVOLVE25:

(* Disclosure: TheCUBE is a paid media partner for Cloudera EVOLVE25. Neither Cloudera Inc., the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE