Dell Technologies predicts governance and sovereignty to dominate AI conversations in 2026

John Roese, Global Chief Technology Officer & Chief AI Officer at Dell Technologies believes that there will be a lot more focus on governance, data management as well as sovereignty in 2026 as companies look to get more value from their AI.

Dell Technologies recently unveiled its annual predictions whereby the vendor believes the rapid acceleration of AI is set to profoundly reengineer the entire fabric of enterprise and industry.

According to John Roese, Global Chief Technology Officer & Chief AI Officer at Dell Technologies, there are five areas in which businesses will need to keep an eye on as they look to make the most of their AI investments.

Specifically, Roese believes that there will be a lot more focus on governance in 2026. He believes that technology and even its use cases are not necessarily going to be successful if companies do not have discipline and governance around how they operate their AI strategy as either an enterprise or a region or a country.

“Our governments are doing their best, but today, a company like Dell has over 1,000 governmental entities around the world with independent AI policies telling us how we should do AI. That is not sustainable. We have been very clear with most governments around the world that while regulation and governance is important, fragmented and chaotic governance rules are not helpful. I think Asia has been more cautious than maybe Europe and other areas. That is good, but we need to all collectively work with our governments to make sure that we have efficient and rational governance frameworks so enterprises trying to do real work have clarity about what the rules are,” said Roese in a media briefing with journalists from APAC.

With governance being the key trending topic of 2026, the first prediction is the need for governance frameworks to be ready for the fast moving ecosystem.

Roese explained that while technology is exciting and even use cases are exciting, the most important thing that must happen in 2026 to accelerate AI in enterprises and governments is clear rules about how will this operate both internally and externally in the markets.

“I spent a lot of time with government officials to help them understand they are an important piece of this process. And I spent more time individually with chief AI officers and CIOs at companies and customers all over Asia helping them primarily establish a governance structure where they could operate efficiently and effectively in getting to production,” said Roese.

The second prediction is the increasing need to have better data management capabilities as it will be the backbone of AI innovation. Roese explained that organizations are still challenged in getting the most out of their data, and this goes down to how they use their data.

“Once you figure out compute and your use case and what you're trying to achieve you realize that data is the center of the AI world. But the data AI systems use is not the same tools, systems, and architectures that we use for traditional systems of record. It is a new layer called a knowledge layer. And so, we think there will be a tremendous amount of activity around defining and implementing that knowledge layer which will include not just new tools but also processes. To make the knowledge layer work it must be accurate and accessible, which means if your data is messy part of building the knowledge layer is cleaning up your data,” said Roese.

As such, he believes that organizations will need to build an explicit data layer called a knowledge layer and that will require new storage, new data protection, new performance levels, new tools and new processes and professional services to make sure it's accurate and accessible.

The third trend is all about the rise of agentic agents. As organizations deploy more agentic AI use cases, these agents will need to be able to talk to each other. Eventually some of these agents will change the way the organization looks and functions.

Roese predicts that agents will not just lead to significant improvement in human productivity but also amplify the skills of the best people in the organization. Because agents can be autonomous and do work, Roese added that there is work that companies fundamentally have chosen not to do because throwing people at it is too expensive.

“When people actually incorporate agents into their organizations, we are predicting that what will happen is a lot of unexpected and maybe not even planned for changes in how the organization works, how agents work with people, how people work with agents, and we will see productivity that is more than just the agents as a tool. Its very presence will bring value to make humans more efficient and to make the non-AI work work better. This is stuff we have already seen, we have seen it repeat, and we expect that as people go on the agentic journey into 2026, they will be surprised about how much more agents do for them than what they anticipated,” said Roese.

The next trend is how AI factories will redefine resiliency and disaster recovery. Dell has over 3,000 customers around the world that are actively building out and deploying enterprise AI and heading to the next big thing. However, Roese pointed out that the industry has yet to figure out how to make AI factories resilient.

“The prediction is we not only have different ways to develop resiliency and a need to have resiliency in our AI factories, but we will have to think about different approaches to what resiliency is. Now, Dell is a leader in this space. We are a leader in cyber recovery, cyber resiliency, cyber vaults, data protection. And so we believe that we are strongly positioned. We believe that all of those technologies will evolve. We are also one of the largest builders of AI factories. And so those two worlds are going to come together dramatically in 2026 and become a major theme. Once you get into production, the next thing you think about is how do I make sure I stay in production? And that is data resiliency for your AI factories,” explained Roese.

The fifth prediction relates back to governance which is on the need for sovereign AI. Roese expects sovereign AI to accelerate national enterprise infrastructure through several ways. This includes having a government build out indigenous AI capacity, data centers that will be used by their industries to make them competitive.

In countries like Singapore, whereby the government works with industries, Roese explained that the government convenes and facilitates the industry to ensure the local companies win, which has been good.

“Once you build a sovereign infrastructure, even if you built it primarily to just run government services or maybe to train big models, once you have large-scale aggregated infrastructure available in your country, there are many other things you can use it for. It is far more valuable than just running your government and providing training infrastructure,” said Roese.

Roese added that as there is more sovereign infrastructure developed, it will enable a lot more use cases than what was initially expected.

“And so sovereign infrastructure is not just for large-scale model training and government services. It creates a foundational capability that will help us with robotics, with agentic, with resiliency, with model management and creation, and ultimately with collaboration of AIs across organizations and around the world. All of that needs infrastructure, and much of that infrastructure is tied to government interest. So we expect sovereign to actually be a much bigger industry than many people are anticipating today,” he said.

The quantum future

Apart from the five prediction trends Roese shared, he also believes that 2026 could see a lot more noise being generated by quantum computing.

“It is moving faster. We've had breakthroughs this year. Quantinium, for instance, has released a 98-qubit fully error-correcting system. IBM has crossed into over 120 qubits in system two. We are seeing advances on the qubit count, but really fully error-correcting qubits, but we're also seeing advancements on software. In fact, we have invested in a company in Singapore that does advanced algorithmic improvement for quantum, and what we found is that those tools are actually reducing the number of qubits necessary to do interesting work. And in fact, today we have examples of algorithms that used to take 10,000 gates to be implemented, that with proper software and optimizations might only take 200. And now we have quantum systems that are rapidly getting into over 100 error-correcting qubits,” said Roese.

Given these innovations, Roese believes that the infrastructure being developed is big enough for the use cases that will create quantum supremacy.

“We are getting closer. The reason that's important is when we get there and we can run viable algorithms on a quantum system, which is not too far away, it will disrupt whatever AI is that day by making it orders of magnitude more efficient and effective than it was the day before,” he concluded.

Interestingly, Roese also stated that 2026 may not see quantum disruption yet, but organizations will need to pay attention to it because there is significant forward progress.