The “Provoke” Moment Is Here for AI and Data Sovereignty

At Davos 2026, the tone around AI quietly but decisively changed. “You have to take control of your destiny,” saidHPE CEO Antonio Neri, speaking to business and government leaders grappling with rising geopolitical risk, regulatory pressure, and accelerating AI adoption. 

Across public and private sessions, the focus may be shifting from AI optimism to AI control, from a race to build capabilities to a race to establish sovereignty over data, infrastructure, and decision-making power. 

When navigating such seismic changes, the greatest risk may not be getting the technology wrong, but waiting too long to act. “If you only react to an exponential shift at the point that you see it, it’s going to be too late. That’s just the nature of exponential change,” said Geoff Tuff and Steven Goldbach, Deloitte strategy consultants and authors of Provoke, speaking on AI & Data Horizons, a podcast hosted by author Michael Gale. 

Davos signaled that AI and data sovereignty have crossed that threshold. The question is no longer if control will matter but when leaders decide to operationalize it. By the time uncertainty gives way to inevitability, options may narrow fast.

Why leaders miss the moment, and the cost of waiting 

According to Tuff, what causes leaders to miss these paradigm shifts is not ignorance but hesitation. Faced with uncertainty, organizations debate and wait for clearer signals—precisely when action matters most. Companies rarely decide to fall behind, according to Tuff. They wait too long to make a decision, and the future arrives faster than their ability to respond.

AI and data sovereignty present a clear case. Most executives already see what’s coming: tighter regulation, rising geopolitical risk, and AI moving from experiments into core operations. 

What’s missing is action. As Kevin Dallas, CEO of EDB, puts it, “Despite the buzz, data and AI sovereignty is still more storytelling than strategy. The winners aren’t louder, they’re just more deliberate.”

According to EDB’s research, the 13% of leaders who have made AI and data sovereignty a mission-critical priority deploy 2x more GenAI and agentic AI applications while reaping 5x ROI, precisely because they have built it into their operating model, aligning data foundations, model governance, and deployment pipelines across the enterprise.

“The cost of waiting isn’t theoretical. Data gets locked into proprietary environments. AI pilots quietly become production dependencies. By the time sovereignty becomes nonnegotiable, the technical and commercial leverage has already shifted away from the enterprise,” said Quais Taraki, CTO of EDB. 

The shift is already underway

What came through clearly at Davos is that AI and data sovereignty have moved out of the realm of future planning and into present-day execution. This is the practical “if-to-when” shift Tuff and Goldbach described: a trend stops being a possibility (if) and becomes an industry-transforming inevitability (when). 

Leaders are no longer debating whether control matters—they’re wrestling with how to regain it after years of convenience-driven choices. As Satya Nadella, CEO of Microsoft, emphasized in a discussion on stage with BlackRock’s Larry Fink: “If you’re not able to embed the tacit knowledge of the firm in a set of weights in a model that you control, by definition, you have no sovereignty. That means you’re leaking enterprise value to some model somewhere.”

In his view, the “phase change” of AI and data sovereignty has already shown up in tangible ways: where regulators are asking not just where data lives but who can access and govern it, boards are asking how dependent core operations are on third-party AI platforms, and security teams are being pulled into AI strategy conversations far earlier than before.

At this stage, sovereignty isn’t a policy discussion. It’s an infrastructure and operating-model decision.

What acting early actually looks like

The leaders who are meeting the moment aren’t waiting for perfect certainty. According to Taraki, they’re making a different set of choices now: In his view, they’re designing AI and data platforms that can run across environments, not inside a single vendor’s walls. They’re treating ownership of data and models as a strategic asset, not an IT preference, and lastly, they’re building agentic AI inside controlled, extensible systems, often on open source foundations. 

Who will provoke the change?

In less than 750 working days, the gravitational pull of AI and data may be felt across enterprises. The only open question is whether leaders will provoke that change now on their own terms—or be overtaken by it.