While the cloud is now critical to modern enterprise, with the tech providing the backbone for 90% of AI initiatives, new research from Kyndryl reveals that most companies have stumbled into their cloud setups without a deliberate strategy.

According ot the IT service provider’s 2025 Cloud Readiness Report, 70% of CEOs admit their current cloud environments were arrived at “by accident, rather than design”, resulting in patchwork systems defined by quick fixes and upgrades, rather than long-term planning. 

However, the lack of direction hasn’t slowed investment, with the report finding that cloud spending has risen by more than 30% on average over the past year, as firms race to implement AI solutions and contend with increasing security demands.

Though the majority (89%) of the 3,700 IT leaders Kyndryl polled said that their cloud investments had made it easier to use AI, 35% still point to AI–cloud integration as a major barrier to achieving ROI.

The growing complexity of AI workloads has pushed most organisations (84%) to adopt multi-cloud strategies, while a number of firms (41%) are choosing to return to on-prem environments, closing the gap between performance and cost by turning to purpose-built infrastructure, like neoclouds optimised for GPU-powered AI tasks.

According to Kyndryl, the rise of agentic AI is driving this rethink around cloud strategy, with sovereignty and security front and centre of their designs.

Many firms are training AI models in public clouds to tap into rapid scalability, then shifting to private environments to meet demands around governance and security as risks heighten.

For example, three-quarters of IT leaders admitted they are concerned about the geopolitical risks of storing and managing data in global cloud environments, and 65% said they have already changed course in response to new data sovereignty rules.

Recommended reading

• Is It Time For a New Sovereign Cloud Code in Europe?
• UK&I SMEs Are Concerned About Data Sovereignty
• UK IT Leaders Shift Cloud Strategy Over Data Sovereignty Concerns

Almost all (91%) of firms said that their cloud infrastructure provides the flexibility to adapt to these new regulations, with Kyndryl arguing that those enterprises approaching cloud as a strategic, intentional capability are best positioned to integrate AI securely, meet regulatory demands, manage costs, and accelerate innovation.

“The gap between a reactive and a deliberate cloud strategy has never been more consequential,” said Nicolas Sekkaki, global cloud practice leader at Kyndryl. 

“With AI demanding seamless data access and governance requirements rapidly evolving, a hybrid cloud model is the differentiator that enables successful AI adoption. Organisations that design for interoperability, trust, and agility will unlock continuous innovation and integrate AI securely at scale.”