Agentic artificial intelligence (AI) will dominate enterprise IT in 2025, as will generative AI (GenAI) from the end of 2022. Behind these two aspects lie many years of traditional AI, based on machine learningadvanced statistical analyses, predictive modeling, etc.
What will 2026 portend? The most likely thesis is that all forms of artificial intelligence (AI) will combine and converge further, with the need for corporate governance determining how this convergence will take shape.
The question of whether the AI bubble The breakup of global stock markets is an interesting issue, but one that is not of much importance to enterprise computing in user organizations. The boom and bust of dotcoms has come and gone, but the Internet has endured and changed every aspect of economic, social, and political life. So this is likely to be the case with AI hyper-growth.
A major problem that Agentic AI specifically posed in 2025, it focused on how organizations manage the identity of agents. It’s hard enough managing the identities of human workers or machine workloads, but autonomous AI systems have proven to be a particular challenge.
Okta Search released in August 2025 indicated the emergence and growth of new security concerns, related to the spread of AI agents and non-human identities. The focus on this topic and the alignment of identity security with agentic AI security seemed to be paying off at the end of the year, as we can see. in its third quarter results.
But many other vendors are turning their attention to the implications of agentic AI for cybersecurity, both from a defender and attacker point of view. ServiceNow’s acquisition of Veza, the first AI-powered identity management platform, and its agreement to acquire Armis, an Israeli AI security company, both announced in December 2025, talk about this trend. And this is just a development.
As agentic AI becomes more secure, the technology is expected to flourish in 2026, bringing more of its vaunted benefits of increasing economic productivity and making potentially more creative and enjoyable jobs – the reason this is potential rather than definitive is that power in organizations is always necessarily contested and it is not a matter of pure technology in the abstract.
The AI hype cycle with the advent of Internet 2.0
However, addressing cybersecurity issues that could further harm enterprise applications is more tactical than strategic in nature for businesses and other organizations.
The question of how the hype about AI was significantly boosted with the advent of Generative AI in 2022 – and its maturation into agentic AI in 2025 – can be productively compared and contrasted with the rise of the Internet in the late 1990s, which deserves more strategic consideration. It’s also a plausible guide to what happens next, in 2026.
In a strong sense, we have been here before. Answers to this comparison question may suggest likely lines of AI development in business in 2026. Computer Weekly asked the question to compare these two phenomena in technology history to a series of top IT industry executives in 2025, during the fourth quarter conference season.
Steve Miranda, executive vice president of Oracle Application Development at Oracle, said ahead of Oracle AI World that he wouldn’t choose customer experience (CX) as the most likely area for AI development: “Around 1998, at Oracle, we had a slogan: ‘The Internet changes everything.’ If you remember, at that time the focus was on e-commerce.”
“Well, e-commerce certainly existed, but the Internet changed music and the way you listen to music. It changed the way you watch – I won’t even say television per se. It changed telecommunications and meetings. It changed everything.”
“We believe that right now AI is going to change everything. There are some charismatic use cases right now, like customer service and customer experience. But when we say it changes everything – and I give examples of digitalization and paying bills and managing (a company’s) ledger – I don’t think anyone appreciates the scale of the impact that this is going to have across the board. We’re at the same point as with the Internet now, except it’s going to be more quickly.”
At SAP TechEd in Berlin, SAP CTO Philipp Herzig said that how enterprise applications develop will be more interesting than how infrastructure and large language models (LLMs) themselves develop. The current rise of AI, he said, is moving much faster than the Internet, adding: “But I always observe that it’s very comparable to the beginning of the Internet. Back then, there was also a lot of focus on the infrastructure plans of the companies that built the switches, routers, etc.”
“Then the discussion moved to ISPs (internet service providers) – the AOLs and telecommunications (companies) that brought the Internet to the home and built distribution to consumers. But, really, the value accumulated in the cloud services we all consume today is in streaming services and cloud software.
“Our belief is that while there is a lot of talk about hardware and around data centers and GPU capacity, for example, that’s not where the end game will be played. The end game will be played in incredible user experiences and native AI software as a service that people can just turn on. We’ll see a lot of new business models that emerge from that.”
It was in Berlin at the beginning of November. In Barcelona later that month, Gerrit Kazmaier, president of products and technology at Workday, told Workday Rising EMEA: “The key thing to understand is that in technology there are cycles, and the first step that always happens is infrastructure innovation. You saw that with Broadcom, and so on, with high-speed Internet. The cloud infrastructure that was built on the availability of bandwidth and the economics that allowed you to do business on the iInternet finally created SaaS (software as a service).
“We’re seeing the same cycle. We’re now at the second stage. We’ve created a very powerful computing infrastructure. We now have the form factor of large language models to essentially use it to create a different type of computation – you might call it computational reasoning.”
“It’s something that didn’t exist before, and now software companies that essentially solve business problems have the ability to take it and change the way business systems are built. Business processes are to a large extent designed around the constraints of human reasoning… But now, with computational reasoning, we have the opportunity to completely rethink.”
At the end of the year, David Richardson, vice president of AgentCore AWS, told Computer Weekly at ReInvent in Las Vegas: “I’m optimistic about the long term, which is important because there will be bubbles and things that happen, ups and downs. The beauty of the Internet is that it has made it possible to have incredibly flexible communication between a wide variety of different computing capabilities, and what people have built from that, no one would have ever predicted. “
“I first used the Internet in 1987, seven years before I used the Web, and it was already fascinating. The other thing that happened with the Internet was that it created a huge ecosystem around building the Internet… (And now) there are so many different participants in the ecosystem who see the opportunity to create value through innovation that it seems like we have many years ahead of us to increase the reasoning and resilience capabilities of these models.
“We’ll continue to find new ways to apply them to problems that we thought we couldn’t solve before or couldn’t solve economically. So that’s my multi-year view of what will happen – 12 months from now, I have no idea and I wouldn’t make a prediction about it.”
What does this fundamental comparison between the AI hype cycle and the dawn of the Internet at the turn of the 20th century tell us about the likely evolution of AI-based business applications in 2026 and beyond? This tells us that a new stage is ready to begin. Classic AI – so machine learning, to put it simply – has now joined generative AI and agentic AI in a way that means enterprise applications will function more like video streaming services that deliver content to consumers algorithmically, and less like the old broadcast television that Gen X grew up with.
Claus Jepsen, CTO of Unit4, suggestively calls this “Ambient ERP“, and his predictions for AI in ERP in 2026 make for interesting reading. In the same way, Steven Webb, UK Director of Technology and Innovation at Capgeminimakes a compelling argument, in an opinion piece on the 2026 forecast for Computer Weekly, that “AI is no longer a supporting technology. It is quickly becoming the operational fabric of modern businesses. We are seeing a decisive shift beyond single-task activities to autonomous, adaptive, self-optimizing systems powered by multi-agent systems.”
This level of autonomy will require technologies and governance processes that have developed since late 2022, when generative AI first emerged, but will need to catch up. Michelle Eisenberg, general counsel at Unit4, has some interesting points to make about creating a AI Governance Frameworkbalancing innovation with a circumspection that extends beyond its origins with this supplier.
But make no mistake, as DXC Technology’s Brad Novak – who has been riding the “mobile, social, cloud and big data” wave, and the rest – told Computer Weekly at the Boomi World Tour 2025 in London: “It’s different, it’s like we all wake up with 20 extra IQ points.” »