For a long time, the economics of on-premises storage were relatively predictable. You could model a refresh cycle, get competitive quotes, and make a reasonably confident CapEx decision. That certainty has gone. Hardware prices have moved sharply in the wrong direction, lead times have stretched from weeks to months, and the structural forces behind both problems show no sign of reversing quickly. For many infrastructure teams, this is starting to shift the conversation away from ownership and towards consumption.
What’s Happening to Hardware Prices
This isn’t a short-term blip. A confluence of factors like AI-driven demand, post-pandemic supply chain restructuring, geopolitical trade constraints, and market cycle dynamics have created conditions that analysts are comparing to a hardware pricing supercycle.
DRAM and NAND flash storage, which hit multi-year price lows in 2023, began climbing sharply in 2024 as the pendulum swung the other way. Hyperscalers with seemingly unlimited appetite for memory and storage have consumed the lion’s share of production capacity, leaving enterprise buyers competing for what remains.
For some memory segments, prices have more than doubled since early 2025. DRAM inventory levels, which sat at a comfortable 13 to 17 weeks of supply in late 2024, had collapsed to just two to four weeks by October 2025. Analysts warn that shortages could persist until 2027, with much of 2026’s production already committed.
Gartner has predicted a combined surge in DRAM and SSD prices of well over 100% by end of year, with shortages now spreading beyond memory into solid state storage, putting further pressure on enterprise IT budgets across the board.
The situation for flash storage is similarly constrained. Manufacturers are increasingly directing production capacity towards higher-margin data centre and AI components, scaling back output of the traditional enterprise storage products that most businesses actually buy. This isn’t a temporary fluctuation, it represents a structural bottleneck that is expected to continue at least through to the end of 2026.
The Lead Time Problem
Price is only part of the equation. Availability has become as significant a planning challenge as cost. Some major HDD manufacturers have sold out hard drive inventory for all of 2026. Smaller enterprise buyers without volume purchasing leverage are being advised to spread purchases across quarters just to average out cost spikes, not because it’s efficient, but because they can’t source what they need at any price in a single procurement window. Procurement has become a calendar problem as much as a budget problem.
For infrastructure teams managing refresh cycles or responding to capacity growth, this introduces a level of operational uncertainty that simply wasn’t part of the calculation before. You can’t plan a storage refresh around a three-to-six-month lead time in the way you could plan one around a three-to-six-week lead time.
Why This Changes the CapEx Calculation
The standard argument for owning your storage infrastructure rests on the premise that capital investment delivers predictable cost over a depreciation cycle. That maths has changed. Purchasing hardware at the top of a supercycle, paying today’s inflated prices for assets that will depreciate over the next three to five years, means absorbing the premium on the way in, rather than benefiting from the cost reduction that typically follows a price peak.
For organisations planning a storage refresh in the current environment, that’s a difficult position to defend to a CFO. You’re not locking in value at a low point. You’re locking in cost at a high one.
The As-a-Service Shift
Against this backdrop, the consumption model starts to look considerably more attractive, not because cloud storage has suddenly become cheaper in absolute terms, but because the comparison point has shifted. When on-premises hardware carries a significant price premium, uncertain availability, and multi-month lead times, the operational and financial flexibility of a managed or as-a-service model starts to close the gap faster than many organisations expected.
The appeal is straightforward: no CapEx exposure at peak hardware prices, no procurement uncertainty, no depreciation risk if the market corrects, and the ability to scale capacity on demand without committing to infrastructure that may need re-evaluating in 12 months.
For organisations running Veeam backup environments, this dynamic is particularly relevant. The cost of scaling on-premises backup storage, whether primary repositories, SOBR extents, or immutable hardened repositories, is directly exposed to the current hardware market. Extending capacity into a managed cloud repository or shifting to a Veeam Cloud Connect arrangement with a provider running at scale, transfers that exposure rather than absorbing it.
What This Means in Practice
The right answer isn’t the same for every organisation. Some will have the procurement leverage and balance sheet position to manage the current market on their own terms. Others will find that the risk profile of on-premises ownership has shifted enough to justify a different approach.
What’s changed is the conversation. Storage as a service used to be evaluated primarily against a stable, predictable hardware cost base. That base has moved and moved significantly. The question isn’t whether storage services are cheap, it’s whether the full cost of ownership, including procurement risk, lead time exposure, and depreciation at inflated prices, still makes the CapEx route the obvious default.
For a growing number of organisations, it doesn’t.
Want to talk through your storage options? Nexstor works with enterprise organisations to design backup and storage architectures that reflect both technical requirements and commercial reality. Get in touch to start the conversation.
Or book a FREE exploratory consultation, with one of our Nexstir experts, at a time that suits you.
