Rising Infrastructure Costs Aren’t A Blip — They’re A Reset

Over the past several months, I’ve gotten a sharp uptick in questions from infrastructure leaders trying to make sense of rising infrastructure costs. The questions are all variations of the same theme: What’s actually driving this? Is it temporary? And what — realistically — can we do about it?

The short answer is that this isn’t a cyclical blip or a transient supply issue. It’s a structural shift in how compute, memory, and storage capacity are produced, allocated, and priced — and it has durable implications for enterprise infrastructure planning.

What’s changed is not just cost. It’s access.

What’s Happening — And Why This Isn’t A Temporary Reset

Enterprise infrastructure markets are being reshaped by an AI‑driven realignment in how capacity is built and allocated. Manufacturing resources, both for the production facilities and for the raw inputs for compute, memory, and storage, are being prioritized for higher-margin AI‑optimized components — GPUs, high‑bandwidth memory, high‑density dynamic random-access memory, and high-performance flash — reducing effective supply for traditional enterprise configurations.

Because these components share underlying fabrication capacity, the supply chain effects stack up. Lower-margin commodity memory and conventional flash storage have been casualties of this compounding, while the shrinking hard disk drive market no longer provides a reliable low-cost fallback for storage. Long-term contracts from hyperscalers and AI infrastructure vendors leave fewer production resources for the conventional compute market, whether it is enterprise servers or end-user laptops and desktops.

The most visible impact is cost. Many enterprise customers are seeing their infrastructure refresh quotes 10–20% above the prices they were getting at the end of last year. This is coupled with a very short-term quote validity window and longer delivery lead times, especially with custom configurations. But the more important story is how long this price hike may endure. Flash memory manufacturers, specifically, have shifted toward margin discipline, passing on higher prices to downstream suppliers. While some new fabs were already being built, new capacity takes years, and anything that breaks down today won’t contribute to supply for 18–24 months.

For the demand side, AI infrastructure order pipelines through 2026 remain high, and new use cases such as self-driving cars will require hundreds of gigabytes of RAM per vehicle, signaling a long-term increase in memory demand. When the supply and demand side stabilizes, expect manufacturers to control production to resist flooding the market and crushing profits. Rather than seeing prices fall to historical norms, the baseline will be higher.

The question that enterprises are grappling with, then, isn’t how long this will last — it’s how to plan and operate effectively when prices are rising and there is no end in sight.

Controlling IT Costs In A Structurally Constrained Market

Enterprises can’t control global supply dynamics, but they can reduce exposure to volatility by changing how they plan, buy, and govern infrastructure. In a supply-constrained market, cost control is less about squeezing unit prices and more about making deliberate choices earlier — around commitments, configurations, and expectations. The organizations adapting best are shifting away from just‑in‑time infrastructure decisions, narrowing the set of configurations they rely on, and placing more emphasis on supply reliability and predictability than on headline discounts alone. Negotiations increasingly focus on clarity, protection, and long-term parts availability, not just price.

Just as importantly, refresh planning is changing. Many enterprises are now assuming higher baseline costs and being more selective about where premium infrastructure is justified while looking harder at utilization and lifecycle discipline to avoid unnecessary growth. Some are even turning to the circular economy, with some infrastructure vendors recertifying used parts for use in production.

The net effect is a more intentional approach to where infrastructure spend increases show up — and where they don’t.

The Bottom Line

Rising infrastructure costs aren’t a temporary anomaly. They’re the byproduct of a structural realignment driven by AI, manufacturer and supplier behavior, and long‑term capacity planning. Enterprises that continue to rely on late‑stage negotiation and just‑in‑time purchasing will feel increasingly exposed. Those that shift earlier, simplify configurations, and treat allocation reliability as a first‑order concern will be far better positioned — not just to manage costs but to keep critical programs moving.

The question is no longer how to get infrastructure cheap — it’s how to get it predictably and plan for the best possible utilization.

If you have questions and you’re a Forrester client, reach out to schedule an inquiry or guidance session.