The AI boom is reshaping the tech landscape in profound ways, and not all of them are welcome for organisations’ budgets. As companies race to deploy generative AI, train large language models, and run inference at scale, the demand for specialised compute has exploded. This surge is forcing hyperscalers and data centre operators to pour hundreds of billions into new infrastructure.
Reports paint a clear picture. Hyperscalers collectively spent hundreds of billions on capital expenditures in 2025 alone, with projections for 2026 pushing toward $600 billion globally – a nearly 40% jump in many cases, largely fuelled by AI build-outs. Those investments are starting to ripple through the entire technology stack.
Major cloud providers and analysts have already warned that the cost of running infrastructure is set to rise, with some providers already adjusting pricing for GPU instances, data transfer, and specialised AI services. Reports point to increases of up to 15% on certain ML workloads, with further changes expected in core services. While some providers are still absorbing the pressure, many have indicated that pass-through increases will be unavoidable as margins tighten.
“iOCO Cloud is taking a different approach. We are not passing these increased infrastructure costs on to our customers. “This isn’t a gesture, it’s a principle. Customers shouldn’t have to pay more every time the cloud infrastructure market shifts. Our job is to manage that complexity for them, not pass the problem on,” says Richard Vester, executive: cloud at iOCO.
When cloud is built right, customers don’t feel market shifts
Vester says the difference lies in how the environment is built and managed. By operating large-scale, multi-tenant cloud platforms, investing in its own infrastructure, and working across multiple cloud providers, iOCO is able to manage cost pressure in ways that many providers simply can’t. “When you run at scale, and you own part of the infrastructure as well as using hyperscalers, you have far more flexibility. You can optimise architecture, shift workloads, invest ahead of demand, and spread cost across environments. That’s how you protect customers from volatility instead of billing them for it,” he says.
“Support costs, licensing, infrastructure – everything is getting more expensive. Simply passing those increases on to customers isn’t a strategy. The strategy is helping them run what they already have more efficiently, making better decisions about where workloads sit, eliminating wasted capacity, and only investing in new infrastructure when there’s a clear operational reason to do it.”
Vester says the current pricing pressure is revealing how easily organisations can be pushed into higher spend without revisiting the fundamentals of their architecture. “When costs go up, the default advice is often to buy more, upgrade, or move to the next platform. That doesn’t always mean it’s the right move. The AI-driven cost surge is actually an opportunity to step back, assess what you’re running today, and make smarter decisions about placement, utilisation, and design before committing to more spend.
“Most organisations already have the platforms they need. What they often lack is the architecture, governance, and optimisation to unlock their full value. When infrastructure costs rise, the answer isn’t panic buying or simply spending more. It’s designing smarter, placing workloads properly, and working with a partner who can see the whole environment. The bottom line is that customers shouldn’t be funding the industry’s growth,” Vester concludes.