How AI Forces Are Spurring a Memory Chip Shortage

Article Highlights:

• In the most extreme cases, contract prices for specific memory commodities like DDR5 have been spiking as much as 100% from one month to the next. It will surprise few people to hear that AI data centers and the complex infrastructure they require are the chief culprits driving these price spikes. 
• HBM is an essential component in AI data infrastructure, delivering the high data transfer rates required for AI models to process the enormous datasets that enable their training and inference capabilities. Because of HBM’s essential utility in the scaling of artificial intelligence infrastructure, AI companies and the data centers they’re building all over the U.S. are driving robust demand for these memory chips.
• It’s important to remember that the memory manufacturing market is significantly smaller than one might expect. Indeed, it’s dominated by just three companies: the aforementioned Micron, Samsung, and SK Hynix. Manufacturing DRAM memory requires highly specialized expertise, and while a larger swatch of companies are capable of assembling memory modules, only a few can effectively execute every step in the manufacturing process. 
• Z2Data’s SCRM platform offers a number of capabilities that can help businesses navigate the rapidly evolving memory market, including access to component prices, inventory, and lead times across all distributors; a proprietary cross engine that helps businesses find crosses to manufacturing part numbers (MPNs) that have experienced significant price increases in 2025; and industry-leading lifecycle forecasting.

It arrived at first gradually, and then all at once. Prices for memory commodities like dynamic random access memory (DRAM) and NAND started to steadily increase in late 2024 and into 2025, as organizations in industries like automotive, consumer electronics, and cloud computing began observing their RAM costs tick upward, again and again, from one month to the next. Then, toward the end of the summer, those relatively modest price increases accelerated, as costs surged from September onward. All told, media outlets were reporting that by the end of the third quarter of 2025, DRAM prices had increased a staggering 172% year-over-year.

In the final months of the year, the situation grew even more dire. Major technology companies, including “hyperscalers” like Amazon Web Services, Microsoft Azure, and Google Cloud, were absorbing price increases as high as 50% on DRAM orders in October and still only receiving partial fulfillments of their original orders from memory chip manufacturers. In the most extreme cases, contract prices for specific memory commodities like DDR5 were spiking as much as 100% from one month to the next. It will surprise few people to hear that AI data centers and the complex infrastructure they require are the chief culprits driving these price spikes. 

But those surging costs are only part of the picture. In this article, we explore the forces behind these increases, including strategic shifts by major memory manufacturers, sustained demand for traditional memory chips, and changes in the geopolitical landscape.

AI’s Appetite Is Triggering a Scarcity in the Memory Market 

To understand how the vast technological ecosystem supporting and advancing artificial intelligence is reverberating through the memory chip market, you need to start with high-bandwidth memory (HBM). HBM is an essential component in AI data infrastructure—and AI accelerators in particular—delivering the high data transfer rates required for AI models to process the enormous datasets that enable their training and inference capabilities. 

To understand how the vast technological ecosystem supporting and advancing artificial intelligence is reverberating through the memory chip market, you need to start with high-bandwidth memory (HBM).

HBM Demand is Reshaping Memory Manufacturer’s Priorities

Because of HBM’s essential utility in the scaling of artificial intelligence infrastructure, AI companies and the data centers they’re building all over the U.S. are driving robust, seemingly inexhaustible demand for the memory chips. This has compelled the world’s foremost memory manufacturers—including Micron, Samsung, and SK Hynix—to divert their production away from DRAM, NAND, and other more traditional memory commodities, and toward making more HBM. The reallocation of manufacturing resources is gradually reducing the supply of and access to more traditional memory commodities. 

The Demand for DRAM, RAM, and NAND Remains Strong

The rub, of course, is that demand for DRAM, RAM, and NAND has not diminished for all the other trillion-dollar industries that rely on memory chips to manufacture their core products. Sectors like automotive, electronics, telecommunications, and cloud computing still require large quantities of traditional memory chips. This has created a stark supply-demand mismatch—and, potentially, an impending memory chip shortage—one in which a steady demand for commodity memory semiconductors is colliding with an increasingly shallow supply pool.

Micron’s recent decision to discontinue its Crucial line of memory chips is highly emblematic of the industry’s larger shift. In a press release from December 3, Micron’s EVP and Chief Business Officer Sumit Sadana explained the thinking behind the chip manufacturer’s pivot. “The AI-driven growth in the data center has led to a surge in demand for memory and storage,” she said. “Micron has made the difficult decision to exit the Crucial consumer business in order to improve supply and support for our larger, strategic customers in faster-growing segments.” Micron is communicating, in no uncertain terms, its decision to prioritize AI customers and their voracious demand for high-bandwidth memory chips over its older lines of memory products. 

Micron’s recent decision to discontinue its Crucial line of memory chips is highly emblematic of the industry’s larger shift.

Few Players In the Memory Market Are Alleviating This Chokepoint

It’s important to remember, here, too, that the memory manufacturing market is significantly smaller than one might expect. Indeed, it’s dominated by just three companies: the aforementioned Micron, Samsung, and SK Hynix. Manufacturing DRAM memory requires highly specialized expertise, and while a larger swatch of companies are capable of assembling memory modules, only a few can effectively execute every step in the manufacturing process. 

As Micron itself explains on its website, “There are only a handful of semiconductor manufacturers with the capability to produce DRAM chips, and they are Micron (Crucial), Samsung, and Hynix.” This means that every decision each of these manufacturers makes regarding how to allocate production capacity has significant ramifications for the entire memory market. Simply put, there are no manufacturers waiting in the wings to step in and seize on the void left when one of these chipmakers shifts their focus to HBM (although there are smaller memory manufacturers, like XYZ, they do not have the resources or production capacity to scale in a way that effectively compensates for the major players shifting to HBM). Instead, supply tightens, prices climb, and we veer closer to a full-fledged memory chip shortage.

A Challenging Geopolitical Landscape Adds Further Obstacles

While the new demands created by organizations investing in AI and the data center infrastructure the technology requires is the key driver of this developing memory storage, it is not the only factor. This year has seen geopolitical tensions bleed into industrial policy, with the Trump administration and Chinese President Xi Jinping each implementing measures—including tariffs and export controls—that have constricted the circulation of semiconductors and other electronic components. Over the course of 2025, this may have gradually impacted the global chip supply, preventing inventories from growing enough to maintain existing price levels. 

This year has seen geopolitical tensions bleed into industrial policy, with the Trump administration and Chinese President Xi Jinping each implementing measures—including tariffs and export controls—that have constricted the circulation of semiconductors and other electronic components.

Timeline of Key 2025 Events in Memory Market 

• September 10: Citigroup issues a warning that the need for DRAM and NAND memory is projected to increase sharply for the fourth quarter of 2025 and into 2026, as memory manufacturers struggle to keep up with exponential growth in demand. Citigroup analysts project a potential supply shortage.

• September 17: Memory prices start creeping upward. DigiTimes publishes data showing contract prices for DRAM and NAND increasing up to 20% for the fourth quarter. The publication attributes the price hikes to the downstream effects of production capacity being diverted to AI infrastructure.

• October 13: Citing industry data from TrendForce, Reuters reports that memory prices for a variety of DRAM chips increased over 170% in the third quarter year-over-year. 

• November 14: Semiconductor Manufacturing International Corporation (SMIC), China’s largest semiconductor foundry, warns that escalating fears around impending shortages in the memory market are leading their customers to hold back on their 2026 orders for other kinds of chips. This suggests that SMIC’s customers are concerned that the memory shortages will force them to cut production on their own products. “People don’t dare place too many orders for the first quarter next year,” SMIC co-CEO Zhao Haijun said during an earnings call. “Because no one knows how many memory (chips) will actually be available — how many phones, cars, or other products it can support.”

• December 1: The GM of TeamGroup, a major supplier of memory modules, warns that a supply shortage for memory chips is imminent, and expects availability to tighten even further in the first half of 2026. The GM said he is anticipating the shortage to last into 2027.

• December 4: Micron announces that it is discontinuing its Crucial line of memory products.

Memory Chip Shortages: A Problem Unlikely to Go Away

With so many developments, disruptions, and seemingly abrupt shifts in the semiconductor manufacturing industry, it can be difficult to distinguish between temporary concerns and serious, enduring issues. It’s challenging, that is, to separate the signal from the noise. But it may be useful to compare the emerging memory chip shortage to another recent crisis—the Nexperia ownership dispute. 

The struggle over ownership of Dutch chipmaker Nexperia seized headlines, in large part due to the novelty of a story featuring two countries aggressively vying for control of a private company. And the Nexperia crisis did reverberate through the semiconductor supply chain, sending shockwaves through the automotive industry and forcing some automakers to modify their operations and production timelines. But the worst of the dispute’s supply chain impacts were also relatively short-lived: the chipmaker eventually resumed its exports out of China and has continued fulfilling its contracts with global customers. (Although there are still lingering complications stemming from the ownership dispute that are impacting global customers.)

It’s unlikely that the supply and demand issues around memory chips, on the other hand, will have a similarly swift resolution. This is a systemic issue in which surging demand for high-bandwidth memory is compelling manufacturers to reallocate fab capacity away from traditional memory. Given the foundational issues at play, it’s no surprise that some industry insiders foresee a shortage lasting into 2027. 

It’s unlikely that the supply and demand issues around memory chips, on the other hand, will have a similarly swift resolution.

Taking Steps to Mitigate the Impact

Z2Data’s SCRM platform offers a number of capabilities that can help businesses navigate the rapidly evolving memory market:

• Z2Data’s Part Risk Manager tool allows companies to see prices, inventory, and lead times across all distributors, including both authorized and independent sellers.
• Z2Data’s Cross Engine can help businesses seek out crosses to manufacturing part numbers (MPNs) that have experienced significant price increases in 2025. 
• Z2Data’s Lifecycle Forecasting gives organizations an accurate, effective estimate of the obsolescence date for their parts, helping them understand the memory chips at the greatest risk of EOL. 

To learn more about Z2Data and how the SCRM platform can help you effectively mitigate this growing supply chain risk, schedule a demo with one of our product experts. You can also contact your Z2Data representative or email us at info@z2data.com.