AI in 2026: fragile data, hybrid clouds & profit race

Technology leaders expect 2026 to be defined less by spectacular advances in artificial intelligence and more by the strain it places on the underlying data, infrastructure and business models that support it.

Senior executives from CTERA and Patmos Hosting say organisations face a twin challenge. They must tackle data governance and infrastructure resilience while investors press for a clearer route to AI monetisation.

Both predict a shift in how enterprises deploy AI. They expect more controlled environments and deeper integration into existing software rather than headline-grabbing standalone tools.

Data foundations

Concerns over the robustness of current digital infrastructure and governance structures are rising as AI adoption increases across sectors such as financial services, healthcare and manufacturing.

"While AI's capabilities seem boundless, the defining story of 2026 won't be whether there's life beyond AI but the urgent need to address the fragile foundations upon which we've built our digital world. The blind rush to adopt AI has exposed a massive governance crisis, revealing that our data is neither as resilient nor as private as we believed.

"The most immediate threat is "shadow AI." In the name of productivity, employees have poured sensitive corporate knowledge into public AI tools, creating an ungoverned, discoverable copy of their company's intellectual property on third-party servers. This quiet data bleed has become an enormous legal and security liability," said Aron Brand, CTO, CTERA.

Brand describes a growing gap between formal corporate policies and the reality of everyday AI use by staff. He highlights the exposure of proprietary information when workers experiment with external chatbots and generative tools.

He also points to structural weaknesses in the cloud ecosystem as AI and data traffic concentrate on a small number of global platforms.

"This internal risk is compounded by an external one: the "single cloud of failure." We traded the internet's decentralized design for a few centralized hyperscalers, leaving our digital infrastructure dangerously brittle. As recent major outages proved, a minor error can now trigger a global shutdown," said Brand.

Hybrid resilience

In response, Brand expects organisations to adjust their infrastructure strategies. He forecasts wider use of hybrid models that spread risk across multiple environments.

"In 2026, these two crises will converge. The smart response will be a strategic reclamation of control. We will see a decisive shift toward hybrid resilience, with organizations replicating data across on-premises and multi-cloud environments to survive inevitable failures. Simultaneously, enterprises will pull AI workloads back inside their own security perimeters, creating private, governed spaces for AI interaction," said Brand.

Brand links AI deployment in 2026 with a broader push for digital sovereignty, as companies reassess where data is stored and how models interact with sensitive information.

"So, is there more to life than AI? Yes. It's building the resilient and sovereign foundation required to deploy it safely. That is the defining challenge of the coming year," said Brand.

AI monetisation

Alongside governance concerns, AI economics remain unsettled. Investment in large data centre projects has surged, but many initiatives are still in early stages of commercial validation.

"Like all successful invaders, AI will eventually become an integrated part of the community. Will this happen in 2026? Not completely, but I think we'll see moves towards that outcome," said Joe Morgan, COO, Patmos Hosting.

Morgan points to high levels of infrastructure spending as providers race to match expectations around AI demand.

"Right now, there are so many gigawatt AI data center builds that promise the infrastructure to meet the "insane" demand predictions for AI. Some of them will fail, and some will succeed, but ALL of them have investors looking for a return on their dollar. Right now, that return is not guaranteed, because nobody really knows how to do it - we're in the crazy part of the hype cycle, and this is why there is so much talk about AI as a bubble," said Morgan.

Embedded tools

Morgan expects AI adoption to progress through gradual integration into existing enterprise software and consumer tools. He anticipates incremental revenue models rather than sudden breakthroughs.

"I don't necessarily think that the bubble will pop, but I do think we will see scale-back in 2026 - and then we will start to see how the industry actually monetizes AI. It won't be scary AGI (yet... hopefully) but it will be a slow integration of LLM AI into the tools and systems we're already using: once you're in the habit of using all those "try our AI for free" tools to write your email or organize your life, they will become paid add-ons to your SaaS, and eventually they'll just be included in the license and you're buying it whether you want it or not," said Morgan.

The comments reflect a broader expectation that AI will sit more deeply inside business infrastructure during 2026. Executives forecast less emphasis on visible novelty and greater scrutiny of resilience, governance and revenue.