Worjd, an anonymous IT manager who posted on Reddit in late February, couldn't believe what he was seeing. Six months earlier, he wrote that he’d ordered a standard server configuration for around $12,000. Now, he was staring down at a quote more than three times that: $45,000.
“I knew prices were high, but this is highway robbery,” he complained.
Such is life lately for many IT leaders. Between AI-driven memory shortages, whipsawing tariff policies, and skyrocketing hardware costs, many CIOs are struggling to keep existing operations running, justify new infrastructure investments, and appease boards of directors who still expect digital transformation to move forward on schedule
The numbers tell the story, with demand for conventional memory and storage products now surging into record territory. Prices for DRAM, the working memory in a computer, climbed 172% year-over-year by the end of Q3 2025, according to Steve McDowell, principal analyst and CEO at NAND Research.
Prices on some DDR5 contracts, a speedier kind of DRAM, also spiked by as much as 100%, McDowell said. Pricing for high-capacity enterprise solid-state drives (SSDs) went up the same way. Storage vendor VDURA reporting a 257% increase for some product categories between mid-2025 and early 2026.
CPU supply has tightened as well. Server CPUs globally are experiencing supply constraints that could push prices up 11% to 15% in 2026, according to a recent Omdia report.
To be sure, today’s supply chain leaders are navigating "a complexity explosion unlike anything we have seen in decades,” Heidi Benko, VP of Product Marketing for Infor Nexus, wrote recently.
The AI Supercycle Nobody Budgeted For
The root causes of all this turbulence are pretty straightforward. Growing Generative AI infrastructure is rapidly increasing the need for top-tier memory and storage, especially High Bandwidth Memory and enterprise-grade NAND.
Very few companies provide these components, namely Samsung, SK Hynix, and Micron. Their production capacity leans toward the most profitable products: high-end chips for feeding GPU clusters at Microsoft, Google, and Amazon. And their earnings have skyrocketed as a result. Micron, for example, is guiding toward 81% gross margins next quarter, according to CNBC.
“There’s a small handful of companies that make memory and flash storage,” NAND Research’s McDowell told Focal Point. “And AI has created an insatiable demand for the high end of those product SKUs. The industry wasn’t ready for this kind of demand on enabling technology.”
Rajiv Ramaswami, CEO of Nutanix, described the issue with CNBC in February.
“There's really no place to go because there's only a handful of people who provide memory," he told CNBC. "We've got to wait for the supply chain to normalize, and that's going to take some time.”
Wall Street has responded in historic ways. SanDisk, spun off from Western Digital as an independent company in early 2025, along with Micron, Samsung and SK Hynix have all seen their share prices hit record highs as investors pile into memory and storage stocks in what analysts expect to be a prolonged "storage supercycle."
This Shortage Could Last Years
For memory manufacturers, this is a golden moment typified by unbelievable profits. For IT professionals buying their products, however, it’s a slow-moving disaster with no quick exit.
"We are not in a boom-or-bust cycle anymore," said Andrea Klein, CEO of the hardware procurement and supply chain advisory firm Rand Technology, in a company blog.
"This is a structural change in technology driven by AI, and the compression and speed of it has never been experienced before. It is unfolding incredibly fast."
Even more concerning is the potential longevity of this restructuring, McDowell, Klein andothers expect it will last at least 18 months, and organizations like Rutgers University warning its staff of the need for “more deliberate planning around procurement, budgeting, vendor selection, and configuration decisions” for the foreseeable future. During Nvidia’s GTC event in March, CNBC reported that some memory chip makers are booking sales contracts that spanned the next three to five years.
"Memory is going to be the biggest constraint,” said Klein.
"There will be no significant new capacity until late 2027, realistically 2028, which means two years of drought. Companies are going to have to decide what they build, what they delay, and what they walk away from."
That’s largely because of how long it takes to bring new semiconductor capacity online.
"The memory and semiconductor guys are building new fabs and doing the things they need to do," said McDowell. "But bringing a new fab online is a three-year construction project."
What CIOs Should Actually Do About It
The situation is difficult, but McDowell said organizations can begin getting a handle on it by rethinking how they use the storage they already have rather than reflexively buying more.
"If I had $100,000 to spend on a storage array two years ago, I could have bought a boatload of storage for that," he said. "Today, it's half that. So, I have to do more with less, and that's where interesting things start to happen."
McDowell recommends intelligent data tiering, placing genuinely hot data on expensive flash storage while moving cooler workloads to hybrid configurations that mix flash with lower-cost spinning disk. That suggests the all-flash-everything approach that dominated enterprise storage conversations for much of the past decade deserves reconsideration. Spinning hard drives, long assumed to be on their way out, are suddenly looking like smart economics again, he said.
"The SSD shortage…who's it good for? It's good for the hard drive guys," McDowell said.
“There are two companies in the world that make hard drives — Western Digital and Seagate, period, full stop. And both of their earnings were through the roof.“
Capacity planning also needs reconsideration, McDowell said. Buying for peak load, the traditional approach, no longer makes financial sense. Buying for current needs and using cloud-based burst capacity to handle overflow keeps capital expenditure lean while preserving the flexibility to scale when necessary, he said.
“A good practice is I might buy my storage or my servers for peak capacity so that I can hit those surges on Black Friday,“ he said. "Maybe I don't do that now. Maybe I buy the capacity I need and then figure out a different way to handle buffer capacity for spikes. Maybe I tier to the cloud.”
The refurbished and secondary hardware markets are other options worth considering. IT procurement firms now routinely advise customers to compare new versus refurbished quotes before committing to a purchase, something that might have been considered a workaround just a year ago.
Multi-vendor sourcing is another structural hedge worth building into a procurement strategy. Rather than relying on a single hardware supplier, certifying servers from multiple OEM vendors means a supply shock at any one of them does not stall a project, according to Nutanix, whose platform is designed to run across hardware from different manufacturers.
For those buying new, McDowell recommends moving fast because prices could continue to climb. At the same time, he notes that maintaining strong vendor relationships is particularly important, especially in constrained markets, because allocation decisions often favor established customers with known histories.
Reports show Dell and NetApp, for instance, have been more aggressive than smaller competitors in locking in component supply and passing pricing guidance to customers early. Where possible, consumption-based models are worth exploring, trading the certainty of a capital purchase for flexibility in a market where prices and availability can shift week to week.
The organizations that emerge strongest from this period will be those that use the pressure to get smarter about how data moves through their systems and where cloud makes more sense than on-premises hardware.
“Scarcity can drive innovation,” McDowell said. “It'll change IT structurally in certain areas, certainly around data management.”
David Rand is a business and technology reporter whose work has appeared in major publications around the world. He specializes in spotting and digging into what’s coming next–and helping executives in organizations of all sizes know what to do about it.
Ken Kaplan contributed to this story. He is Editor in Chief for The Forecast by Nutanix. Find him on X @kenekaplan and LinkedIn.
© 2026 Nutanix, Inc. All rights reserved. For additional information and important legal disclaimers, please go here.
Related Articles
