Top Technology Trends in 2025 and 2026. Part 1
As 2025 draws to a close, it is time to look back at this year’s major developments in technology and consider what lies ahead for 2026. In this first part of our podcast, technology writer Mark Hillary and industry analyst Peter Ryan reflect on the predictions they made last year, examine how those themes unfolded across global enterprises, and highlight the developments that became central to decision makers in 2025.
Mark:
I think we made about half a dozen predictions this time last year, and those were around agentic AI, AI governance, cybersecurity, and disinformation. Those four definitely became very important themes this year. So out of the six predictions that I saw in my notes, those four clearly landed as major trends.
We also talked about cloud and I think the problem was that we kept cloud too general, just saying that cloud computing would be an important trend. What really happened was that cloud became subsumed by AI. AI delivered across the cloud became an important trend, but saying that cloud alone would be the trend wasn’t quite accurate.
We discussed analytics as well, and again, I think what we didn’t predict was how important AI would become to analytics. Automating analytics and business information became a key trend, rather than simply saying that control of business metrics would be important. There were definitely four out of six predictions that landed. And for the other two [security and disinformation], I think the trends were real, but maybe needed more precision in how we described them.
Peter:
I would 100 percent agree with Mark. In the case of cloud and analytics, I would give those predictions more of a soft landing.
Both Mark and myself travel to different trade shows and conferences. The extent to which cloud and analytics are on the minds of executives and decision makers in enterprise technology and infrastructure management is huge. Mark makes a very good point.
The key around cloud and around analytics is the extent to which AI can be supported by cloud or virtualized infrastructure, and in the case of analytics, the extent to which analytics can be powered or accelerated by AI. The reality is that these tools and these elements are still very much on the minds of decision makers. Going back to the four predictions Mark outlined, I think we were pretty close.
The buzzword of 2025, the past 12 months, has been agentic AI, copilot, agent assist – all these capabilities that make workers’ lives easier and more efficient. That has been a major factor in how technology tools have been procured in 2025 and will likely continue into 2026.
I would add a point about cybersecurity. One of the things we’re tracking in our research, including research coming out in the next couple of weeks for 2026, is that cybersecurity is omni important right now. Decision makers are reinforcing their information security and cybersecurity plays a major role in oversight operations. This becomes more important every year. It’s not something we’ve seen de-emphasized at all over the past decade. And I think that in 2025, this was absolutely the case.
Mark:
If we’re talking about trends moving from this year into next, I would say cybersecurity will continue to be one of the most important trends. It will develop into a need not only to secure existing information, but also to understand the provenance of the information we use.
This especially applies to training LLMs or small language models. Where is the data coming from that we train these models with? That type of cybersecurity is going to become extremely important because it helps answer questions around hallucinations. We’ve seen several problems this year.
A good example is the toy market. If you look at recent U.S. media coverage, they’ve been urging parents not to buy AI-enabled toys because there have been several cases where toys designed to learn about your child, understand their name, preferences, and dislikes so they can have conversations, were behaving unpredictably. They found that if you speak inappropriately to these toys, the toy will learn and echo back these inappropriate conversations.
There is this need for security to also include the idea of provenance. Where is the training data coming from? How are we training AI models to ensure they work correctly?