AI Infrastructure Boom Tightens IT Supply and Budgets

By Amara Osei

Memory and storage price increases have affected the IT budgets of 87% of respondents in an Omdia survey. Nearly three-quarters said shortages were slowing their ability to deploy AI infrastructure on premises, while 80% were looking for greater infrastructure efficiency.

Infrastructure suppliers are benefiting from the AI buildout, but the same capacity squeeze makes it harder for CIOs to refresh everyday systems. Omdia principal analyst Simon Robinson discussed the survey as organizations confront new assumptions about equipment availability, delivery dates and cost.

Where is the squeeze coming from?

AI’s rapid expansion of data centers is intensifying shortages and pushing manufacturers toward high-volume, higher-margin products. Hyperscalers already bought enormous amounts of infrastructure before the AI boom, and supply chains were already under pressure. The result reaches beyond AI builders: rising prices and uncertain delivery schedules complicate ordinary IT purchases and can change where organizations run workloads.

Shane Snider’s analysis of infrastructure suppliers’ earnings shows the scale of demand:

• AMD’s data center revenue nearly doubled year over year.
• Arista’s networking revenue rose 58.5%.
• Eaton’s electrical equipment backlog grew 46%.

These figures cover different markets and cannot be attributed entirely to AI. Together, however, they show demand reaching processors, networking and power infrastructure while capacity constraints build.

Stephen Sopko, semiconductor and deep tech practice lead at HyperFrame Research, told Snider: “For silicon delivery, the bottleneck sequence is still memory, then packaging, then everything else.”

The shortages may not be new, but their scale changes assumptions about what equipment will be available, when it will arrive and what it will cost.

Takeaway: AI did not start every shortage, but its data-center expansion now competes for capacity and raises costs across enterprise technology purchasing.

How are CIOs changing purchases and deployment?

Some IT organizations are delaying nonessential purchases, keeping equipment longer and using existing capacity more efficiently. Others are reconsidering where they run workloads. The pressure now extends beyond data-center equipment.

At educational testing organization ETS, Lenovo laptop prices rose about 25%. CIO Steve Santana told InformationWeek that ETS responded by extending its laptop refresh cycle from four years to six, adding two years. The change eases immediate budget pressure but can increase maintenance costs and operational risks.

Public cloud can provide another route around a hardware purchase or delivery delay. Some organizations use cloud storage as a temporary response to shortages, while others are reconsidering broader public-cloud adoption. It does not guarantee savings.

Jeff Groom, director of engineering, AI, at Acre Security, told TechTarget’s Everett Bishop: “Cloud and SaaS providers can absorb some of the supply-side shock, but they don’t remove the cost.”

Beth Pariseau’s reporting on private AI shows why workload placement is not a one-way decision. Enterprises retain reasons to keep particular systems in-house, including cost, control, security and the flexibility to run models on their own terms. A shortage changes that calculation but does not settle the choice for every workload.

The effects extend beyond IT budgets. A delayed infrastructure purchase can postpone modernization, leave employees working with aging equipment and force customer-facing improvements to wait.

Takeaway: CIOs have several ways to manage the squeeze, but every delay or cloud shift exchanges one cost or risk for another.

How should CIOs plan for costs they cannot control?

CIOs already balance AI investments against inflation, security, aging systems and modernization needs. Forrester analyst Brian Hopkins told TechTarget’s Kinza Yasar that enterprise software spending is growing more slowly than inflation, eroding purchasing power as other demands increase.

No IT department has an unlimited budget. When the numbers cannot support every planned investment, the choice may be the least bad option rather than a good one.

The organization’s own data should provide the starting point. CIOs need to document the condition of existing infrastructure, how long it can reasonably last, the cost of maintaining or replacing it, and which business operations depend on it. Those findings should sit alongside forecasts for prices, availability, delivery schedules and other market pressures.

Rather than abandon refresh cycles, IT leaders can use them to build a tiered set of options:

• Purchases or projects that cannot be postponed.
• Purchases or projects that can be adjusted.
• The operational and business risks created by each delay.

Business leaders should join the process early, with enough information to understand the choices and set priorities. Some organizations may bring purchases forward. Others may extend equipment lifecycles, change vendors or move selected workloads to public or hybrid cloud.

Takeaway: CIOs cannot control the global supply chain, but they can rank business needs and decide which consequences the organization can afford.