Service-based IT & agentic AI to reshape 2026 spend

Technology executives expect 2026 to bring a decisive shift in how enterprises consume IT, run infrastructure and operationalise artificial intelligence, with service-based models, automation and real-time data platforms set to dominate spending agendas.

Senior leaders from Hitachi Vantara and Extreme Networks predict that consumption-based IT, agentic automation and more specialised AI models will shape strategies as organisations attempt to scale digital workloads without expanding headcount.

They also highlight growing pressure on networks, storage and data centres as businesses seek to move AI from pilots into day-to-day operations.

Service models

Infrastructure purchasing patterns are expected to tilt further towards operating expenditure in 2026, with subscription and pay-per-use approaches displacing traditional ownership.

"Consumption-based and service-driven infrastructure models will become the primary way enterprises deploy and manage IT. Traditional CapEx purchases and fixed leases will continue to decline as organisations favour predictable, flexible "as-a-service" models with pay-per-use pricing to reduce upfront costs, scale quickly and improve ROI," said Jeb Horton, Senior Vice President of Global Services, Hitachi Vantara.

Horton expects this shift to alter how CIOs and finance teams plan for infrastructure, with predictable billing and flexibility becoming as important as technical specifications.

Automation impact

Horton does not expect widespread job losses in IT from automation next year, despite the rapid take-up of AI tools.

"Despite common fears, AI and automation will not dramatically reduce IT staffing in 2026. While these tools are dramatically boosting productivity, the demands for IT resources are also surging. As a result, companies will manage two to four times more workloads with roughly the same number of people, realising tremendous efficiency gains without significant headcount reductions," said Horton.

The focus for many firms is likely to be on stretching existing teams across larger estates of applications, data and infrastructure, while keeping operational risk under control.

Managed services

As environments grow more complex, Horton expects broader use of managed services beyond public cloud platforms.

"Customers will increasingly seek on-prem, hybrid and multi-vendor managed services rather than just employing hyperscalers for their infrastructure, compute and storage needs. Providers that can manage heterogeneous environments, including competitors' hardware, will have a competitive edge," said Horton.

This suggests a growing role for service providers that can integrate and operate assets across data centres, private clouds and multiple public clouds, rather than focusing on a single platform stack.

Real-time infrastructure

Enterprises deploying AI into customer-facing and operational workflows are expected to prioritise infrastructure that can process data continuously and at low latency.

"With AI, we live in a real-time world. To enable faster decision-making and automated responses - especially in customer-facing interactions - enterprises will need real-time infrastructure, which is the technology foundation that captures, processes and acts on data the moment events occur. This combines fast compute, high-speed networking and powerful storage to support instant decision loops and responsive operations. In 2026, this capability will move from innovation to expectation," said Horton.

Horton links this to AI deployment at scale, where infrastructure performance and data flow become direct constraints on what systems can deliver.

Data foundations

Generative AI projects are also likely to expose long-standing weaknesses in data management, Horton said, pushing more investment into data quality and governance.

"Generative AI growth will continue to strain networks, storage and compute, forcing upgrades in data availability, curation and throughput. But AI success will depend less on algorithms and more on data quality, organisation and curation, echoing long-standing analytics challenges. Without that, downstream outcomes will falter," said Horton.

"In 2026, organisations are likely to see AI and network automation achieve levels of autonomy and agency never seen before. With agentic automation, we can expect a major shift toward proactive and predictive AI capabilities, enabling systems to anticipate needs, make intelligent decisions, and optimise operations in ways that go beyond anything we've seen so far," said Markus Nispel, Head of AI Engineering and EMEA CTO, Extreme Networks.

AI as priority

Nispel expects AI to remain central to corporate strategy, with a shift from experimentation to targeted, commercial outcomes and resilience against disruption.

"We will see AI continue to be a top priority for businesses in 2026 as a general-purpose technology. Beyond driving productivity gains, companies will increasingly focus on leveraging AI to transform, and potentially disrupt, their own business models before competitors or new market entrants do. That means focusing on high-value use cases that deliver measurable, real-world results. Robust, scalable infrastructure will be critical to enable this, as businesses cannot unlock the full potential of AI without a strong foundation. At the same time, I expect we'll see more attention on applied AI, making sure these technologies actually work in practice and drive tangible business value," said Nispel.

According to Nispel, this will require closer attention to the interplay between AI workloads and network architectures, as systems take on more autonomous decision-making.

Next-wave AI

Beyond today's large language models, Nispel expects discussion in 2026 to concentrate on new approaches and their infrastructure demands.

"The top technologies people are likely to be talking about in 2026 will go beyond today's LLMs. We'll see a lot of focus on increasing the capabilities AI can provide through multi-agent systems, leveraging small and domain-specific language models, and exploring how to achieve artificial general intelligence with a new category of AI, world models. Another big conversation will be around the impact of these advancements on global GPU infrastructure, including how data centres will need to scale to support these next-generation AI systems," said Nispel.