The Independent was not involved in the creation of this sponsored content.

Is the agentic organization being built the right way, the sovereign way? This is a choice that is still open today, though the window of opportunity may not stay open for long.

Ninety-five percent of enterprises want to run their own AI and data platforms by the end of 2028, according to recent research from EnterpriseDB (EDB). That is a fundamentally different business model and a fundamental shift in operating expenditure. The research demonstrates a stark gap: only 13% of enterprises are already there today — and they are seeing five times the return on investment of their peers and running twice as many agents in full production, often across 10 or more business functions. If this is the shape of the enterprise in 2031, the real question is how to get there, because that choice will radically shape how to design an organisation for success.

Successful enterprises often position sovereignty as the foundation. Being able to access, build, deploy, orchestrate, and learn from all of an organisation’s data, holistically underpins everything else. That pattern can hold everywhere, from Japan to the U.S., the UAE, Brazil, and India: Sovereignty is a choice, not a barrier to success.

It may also demand a shift in technology philosophy.

“Openness and agility often go together because nobody can control the future, only how they respond to it. That means bringing every layer together: governance at the data layer, unified data, runtime, and deployment systems. AI, data, and infrastructure become, in effect, the operating system for AI success,” says Kevin Dallas, CEO of EDB

As the Nobel-winning economist Daniel Kahneman has observed, hindsight makes the past look far more predictable than it ever really was. According to Dallas, the success of the leading 13% offers a template for what’s coming: These are enterprises that are deeply agentic by design, that treat data and AI as competitive moats, and that see the value of data across every function as a competitive advantage. The critical question centers on how these shifts will reshape the operating costs of a $1B enterprise between 2026 and 2031.

“Sovereignty isn’t about slowing down to be safe,” says Dallas. “It’s about designing your AI and data foundation so you can move faster, with control, at the exact moment agentic AI scales across your business.”

**EDB says sovereign control could be 40%+ of OpEx by 2031 **

A typical $1B-revenue enterprise today might spend upwards of 15% of revenue on technology and cloud. EDB CEO Kevin Dallas envisions that by 2031, that same company could be spending 40% or more of its revenue on AI and data infrastructure — much of it in the cloud — running perhaps 60,000 agents, with every business unit measured on how well its AI and data serve as a competitive moat.

“Enterprises often underestimate how much of their cost base will shift toward AI and data infrastructure by 2031,” Dallas says. “The winners are the ones treating that shift as a design decision today, not a budget surprise five years from now.”

Aspiration has to meet operational reality. Sovereign controls will need to sit across a significant amount of OpEx, in engineering, HR, sales, and everywhere in between, according to Dallas. Every department will become its own AI and data competitive moat. But without visibility into how value is created and how effective each function is at building genuine differentiation, enterprises may not be able to see whether their data is truly unified; whether governance is built into the data layer their agents rely on every second of the day; or whether those agents can be deployed wherever they need to act, learn, and recalibrate in near real time.

“By 2031, the most successful enterprises across every sector — from banking and finance to telco, healthcare and retail — will see their AI and data as one operating system across the enterprise: their competitive and sovereign asset,” says Dallas.

A sovereign AI operating system is a design choice. Get it wrong, and enterprises may bear the costs of aspiration without the rewards of the right design. Done properly, AI can stop being confined to one or two functions and become a competitive moat everywhere, from faster marketing ROI to levels of sales personalisation unthinkable a year ago to customer service that anticipates a need before the customer realises they have one.

The first step is getting all of an organisation’s data and agents onto a secure, agile, unified platform — one with sovereign control and the observability to see, learn, and adapt. By 2031, that might become standard practice.

“The only real choice enterprises may have now is whether to get there early, like the leading 13%, or follow much later,” says Dallas.