Companies today are operating amid intense instability and erosion of trust among their stakeholders and customers, threatening their long-term growth outlook. Geopolitical tensions have disrupted global supply chains and threatened data centers that power the digital economy. Regulators are imposing new rules on technology that barely existed two years ago. Meanwhile, the “tokenocalypse” has reshaped the economics of AI.
From government policy debates in the EU to boardrooms around the world, the conversation is moving from globalization to “sovereignty” as a more sustainable operating model. Every leader today needs to consider how to hedge risk, control cost, respond to unexpected events with agility, and serve their customers without disruption.
For CIOs, this means prioritizing resilience and adaptability, without stifling innovation. They’re reconsidering where they host and run applications and where they should store their data to mitigate vulnerabilities. They are also reconsidering which AI vendors are less susceptible to global instability.
In practice, this means distributing workloads across multiple regions and jurisdictions. They are avoiding deep coupling to a single cloud provider’s proprietary services, and building AI model optionality into their architecture. This is the new sovereign enterprise. It is one that strengthens its operations with autonomy and flexibility and exercises fine-tuned control over its AI investments, so it can serve its global customers without friction and chart new paths to growth.
CONTROL AS A STRATEGIC PRIORITY
Imagine a mission-critical workflow suddenly failing because an AI agent no longer behaves the same way after an LLM update. Or that the SaaS you use to run a core business process has embedded a proprietary AI model deep into its architecture. Your data, operations, and future capabilities can suddenly become dependent on a vendor you never chose.
These are not theoretical scenarios. In conversations I have with customers, such as the Portuguese health services provider Luz Saúde, the Red Cross in the Netherlands, or government agencies around the world, I am constantly reminded that the systems enterprises run today are part of the infrastructure of modern societies. It raises a fundamental question: When AI is embedded throughout these organizations, who is in control?
CEOs are in a race to reinvent their companies around AI, but first, they should be asking their CIO how they plan to build autonomy and agility into their enterprise systems. Otherwise, they risk surrendering to AI control of core operations they don’t fully understand and they give up leverage in future vendor negotiations.
THE STRATEGIC RISK OF AI DEPENDENCY
When I started my career in the early 90s, the tech industry was already being defined by a small number of players. The PC revolution was built around a few operating systems. Cloud platforms defined the last decade of enterprise software. AI introduces a new level of concentration risk. When only a few companies can afford the chips, data centers, and training pipelines to build the leading models, power naturally concentrates in a small group.
Recent events show how fragile that concentration can be. Legal and regulatory disputes are emerging between governments and major AI vendors over security, ethics, and supply chain risks. If an enterprise builds critical systems on a single AI provider, it inherits that provider’s roadmap, pricing model, geopolitical constraints, and regulatory exposure.
No CIO wants to explain to their board why a strategic system stopped working because a model provider changed policy. Or they changed pricing or availability. Tech leaders learned this lesson before with cloud and SaaS lock-in. They shouldn’t make the same mistake with AI.
FROM AI TOOLS TO AGENTIC ECOSYSTEMS
The changing nature of AI will demand interoperability over monolithic systems. What began as models generating text or code snippets has, in a matter of months, evolved into agentic systems that coordinate work across enterprise applications, data, and business processes.
Some organizations are already operating thousands of specialized agents that collaborate across business processes to analyze data and automate decisions. The agents also interact with core enterprise systems. Those agents will not always run on the same AI model. Different models will be better at different tasks. Some will run locally. Some will run in the cloud. Some will be proprietary, while others will be open-weight. To leverage this diversity, organizations must design for model flexibility and choose the best approach for each use case.
This shift fundamentally changes how enterprise systems will operate. In the past, applications were the unit of systems architecture. In the agentic era, orchestration is the new center of gravity. Companies need a way to coordinate multiple applications and agents and integrate with enterprise data, while maintaining governance over autonomous systems acting on their behalf.
ORCHESTRATE INDEPENDENCE
A new platform orchestration layer is emerging as the essential foundation for a sovereign enterprise. Designed for interoperability between AI models, agents, and existing applications, the orchestration layer will give CIOs a master control panel for managing their entire technology estate.
Instead of committing to a single AI provider, an agentic systems platform abstracts the underlying models. It allows developers to swap, combine, or replace them without rebuilding the surrounding system.
With this approach, models become another interchangeable component. Agents coordinate workflows, and the agentic systems platform becomes the operating environment for AI. This ensures governance, cost control, security, and context across the full lifecycle.
TOWARD A TRUSTED SOVEREIGN ENTERPRISE
The AI industry is moving at an incredible speed. New models appear every few weeks, and capabilities improve dramatically with each generation. In this environment, locking an organization into a single provider becomes a liability.
The enterprises that succeed in the coming decade will design architectures that allow them to adopt the best available technology at any given moment. That means creating systems where teams can switch models, orchestrate agents, and keep enterprise knowledge under the organization’s control.
Our economic future depends on companies being able to provide critical products and services regardless of geopolitical tensions and changing regulatory environments. The sovereign enterprises that have derisked their technology infrastructure and achieved AI autonomy will be well positioned to earn stakeholder trust and grow in an increasingly uncertain world.
Woodson Martin is CEO of OutSystems.