The United States’ Accelerating Memory Chip Crunch: Implications for Tech and AI Leadership

Over the past year, AI data center expansion has heightened demand in the United States for the most advanced high-bandwidth memory (HBM) chips, creating adverse effects across the broader memory chip ecosystem. As memory chip suppliers prioritize selling to large tech firms purchasing HBM for data centers, supplies of conventional random-access memory (RAM) chips required to produce high-demand consumer electronics such as cell phones, laptops, game consoles, and cars are becoming increasingly scarce. The resulting “RAMageddon” is driving up prices for the memory chips feeding the consumer digital economy, and in turn, raising costs for consumers.

With only a small number of memory chip suppliers—specifically Micron, SK Hynix, and Samsung—serving the U.S. market, and with new facilities essential to increased capacity several years from coming online, concerns are growing. Large technology firms, consumer electronics companies, and legislators alike are calling for government intervention to address the shortage and expand supply—including by looking to Chinese-manufactured chips. Meanwhile, the three memory chip suppliers have cautioned that government intervention could produce adverse effects on the market over time.

Q1: Why is the technology community concerned about a so-called “RAMageddon?”

A1: U.S. tech headlines this year have sounded alarm bells due to a supply-demand crunch for memory chips. On one end of the spectrum, there is intensifying demand for the most advanced HBM chips powering the AI data centers whose footprint is rapidly expanding across the United States; these in turn generate the compute needed to run the most advanced AI models. By some estimates, between 50 percent and 70 percent of all memory chips produced this year will be consumed by data centers. At the same time, a growing number of consumer technology companies and policy experts are sounding the alarm regarding an adverse effect of this skyrocketing AI chip demand: It is simultaneously creating severe supply shortages of less-sophisticated conventional RAM chips required to produce high-demand consumer electronics such as cell phones, laptops, game consoles, and cars. This chip scarcity is driving up costs that companies are already passing on to consumers in the form of higher prices—causing a so-called “RAMageddon.”

Over the short term, the prognosis is challenging. The United States relies on just three firms—two Korean, SK Hynix and Samsung, and one American, Micron—to produce the needed consumer memory chips. All three companies are highly constrained in generating more immediate capacity. SK Hynix’s CEO recently warned that the broader memory chip industry was in for its “worst” shortage in history, contending that even with new facilities built, demand would outpace supply by 2030. Intel’s CEO expressed a somewhat less dire view when he stated earlier this year that “there’s no relief [in sight] until 2028,” as the three companies—which together control 95 percent of global supply—will need to build new facilities to increase production. Micron is constructing two new Idaho facilities that are not expected to open until at least 2027. In the coming years, Samsung and SK Hynix also plan to spend upward of $500 billion on two new South Korean facilities each to increase production.

Q2: What are the different categories of memory chips? How is the memory chip shortage affecting the broader U.S. technology sector?

A2: To explain the problem, it is necessary to define several key terms. RAM is the catch-all term for short-term memory chips. Dynamic Random Access Memory (DRAM) more specifically describes the conventional type of RAM used in laptops, personal computers, phones, and many other consumer electronics. DRAM prices have historically decreased over time, but recently that trajectory has reversed as costs have trended up due to supply constraints. These constraints are driven by hyperscalers’ demand for HBM, a product composed of stacked and connected DRAM chips that deliver the higher bandwidth and lower energy consumption needed for AI. (By contrast, NAND flash memory provides long-term data storage even when a device is powered off.)

As the urgency of the term “RAMageddon” suggests, the intensifying memory chip shortage has broad implications for the U.S. tech sector—with impacts varying depending on where companies sit in the ecosystem. For the chip suppliers and their shareholders, the outlook is comparatively rosy. While high prices for HBM are not desirable, large tech firms (including hyperscalers) are nonetheless able to afford them and secure additional future supply. For consumer electronics firms and their customers, the impact is universally negative: higher price tags on consumer electronics such as smartphones and laptops, for which DRAM chips account for up to 30 percent of input costs.

Chip suppliers and their shareholders benefit from the current chip shortage. The U.S. and South Korean companies have publicly addressed the current shortages and inherent trade-offs, and their posture in the current market appears uniform. SK Hynix, Samsung, and Micron acknowledge their corporate strategies favor HBM chip production to fuel AI’s explosive growth. According to industry reports, SK Hynix—as Nvidia’s primary supplier—has shifted most strongly toward HBM production, at the expense of DRAM, and has seen its own profit margins increase accordingly. The firm plans to build new facilities in the United States (with some support from CHIPS Act funds) and South Korea, but these are unlikely to alleviate shortages until at least late 2027. Samsung is the leading global memory chip producer, and it also benefits from supplying its own smartphone models with needed chips, while also selling to leading AI chip designers Nvidia and AMD. Finally, Micron, the only U.S.-based memory chip manufacturer, is critical to the U.S. government’s strategy to expand and prioritize chip production domestically. The 2022 CHIPS Act created a $50 billion fund, with $39 billion set aside for manufacturing incentives. The Commerce Department later utilized these authorities to award Micron more than $6 billion for two new facilities in Idaho and one in New York (SK Hynix also stands to receive more than $450 million in CHIPS funds). While Micron reports that its projects are actively underway, it will be years before any are producing new chips in the United States. And, like its competitors, Micron is meanwhile prioritizing HBM production.

Hyperscalers—or large AI firms and cloud providers such as Google, Amazon, Microsoft, and Meta—are simultaneously the greatest drivers of advanced chip demand and the most capable of paying higher costs today while storing up future capacity. Indeed, hyperscalers today are spending massive amounts on AI infrastructure—upward of $100 billion each—all while experiencing preferential treatment in the constrained chip market, as they receive priority access to the HBM chips they need for AI data centers. Consequently, the South Korean and U.S. chipmakers have shifted productivity toward HBM at the expense of the DRAM chips needed as inputs for consumer electronics.

On the other end of the spectrum, consumer electronics and other technology companies (including telecoms and automobiles) are directly experiencing the crunch generated by soaring hyperscaler demand for HBM chips, which is in turn driving up costs for more scarce DRAM chips. Companies from Dell to Lenovo, Apple, and Microsoft have directly tied increased consumer product prices to DRAM chip shortages. For example, heading into 2026, Dell expected to raise personal computer prices between 15 and 20 percent; Lenovo indicated it would raise prices on laptops; and, in explaining price increases for new products in June, Apple CEO Tim Cook claimed the company tried “to shield . . . [its] customers from the increases, but the situation has become unsustainable.” Beyond phones and laptops, other tech such as telecoms equipment and cars are similarly reliant on these memory chips and are seeing component costs rise. Swedish telecoms maker Ericsson, for example, recently reported that surging AI chip demand, at the expense of conventional DRAM, raised costs for its base stations. Carmakers are particularly affected and locked into the existing Western chip providers, as they (and other industrial technology sectors) must account for the substantial time it takes to qualify component parts.

Q3: What are the likely downstream impacts of the current memory chip shortage?

A3: Concerns over widespread impacts fueled by the consumer memory chip shortage are reaching a fever pitch, with industry stakeholders and members of Congress pressing the U.S. government to come up with solutions, from interventionist to light-touch. Recently, nine industry associations representing a wide range of consumer electronics sectors sent a letter to Treasury Secretary Scott Bessent and Commerce Secretary Howard Lutnick decrying the “urgent imbalance in the market for memory chips that could lead to significant and sustained near-term price increases for American households and disrupt critical U.S. supply chains.” The consumer product manufacturers argued that reduced memory supply is already generating higher prices for companies and customers “for a broad range of everyday consumer electronics and information technology products;” causing more expensive internet and telecoms infrastructure costs; and leading to disruptions to car and medical device manufacturing, along with a range of other concerns. Outside industry, legislators such as Senator Bernie Moreno of Ohio are pressing Secretary Lutnick on the United States’ vulnerability to the chip shortage given reliance on just one U.S.-based manufacturer, Micron. His letter emphasized auto industry impacts and pressed the Trump administration to prioritize meeting domestic RAM demand in its engagements with Micron, SK Hynix, and Samsung.

How will customer demand change over time? It is likely too soon to say. In the short term, electronics retailer Best Buy recently reported that customers currently appear undeterred by rising costs and are still buying needed electronics such as laptops. But other analysts expected shipments on personal computers and laptops to fall in 2026. Higher costs could also slow the adoption of AI-enabled computers through 2027 and sharply affect the entry-level computer market. Over time, many concerns remain—including among executives at the National Retail Federation—that consumers may be less likely to replace electronics if price hikes continue to rise unabated.

Q4: What options have industry and government put on the table to remedy the current shortage?

A4: To address the crisis, consumer electronics–facing industry organizations have urged the Treasury and Commerce secretaries to take urgent actions: first, generate greater U.S. and allied country memory chip capacity; second, work toward trade deals that facilitate supply chain resilience; third, ensure existing capacity is distributed to consumer-facing firms; fourth, reexamine CHIPS Act funding to find ways to relieve the shortage; and fifth, address regulatory barriers that limit global supply. On generating additional capacity, one possibility is for the existing memory chip makers to establish partnerships with other chipmakers who have facilities that could be converted to produce them.

Apple has taken the most visible role in advocating for limited use of Chinese memory chips, while lawmakers have signaled that other, unidentified U.S. technology companies are also exploring the potential for this solution to help relieve the supply crisis. (And the companies are not alone, with some expert analysts arguing that the United States should seriously consider this option). Apple has argued that the memory crunch impacting nonsensitive consumer electronics makers could be alleviated in the short term should firms be able to buy Chinese-manufactured chips. Today, two main manufacturers comprise China’s memory chip ecosystem: CXMT (ChanXin Memory Technologies) and YMTC (Yangtze Memory Technologies). CXMT is already the fourth-largest global DRAM manufacturer and a supplier to Huawei, and NAND flash memory supplier YMTC’s market share is rising.

However, the two firms currently reside on the Pentagon’s 1260H List, which is reserved for companies the department determines are linked to China’s military—although both were briefly removed in February for unexplained reasons and then reinstated. CXMT is simultaneously subject to EAR 744.23, while YMTC is on the Commerce Department’s Entity List, making it challenging to produce these chips without obtaining an export license. For Apple, or any other consumer electronics manufacturers, one proposed way to expand access to needed chips, while also mitigating national security concerns in the United States—including supply chain security and cybersecurity—could be by seeking to restrict products with Chinese-made chips to be exclusively sold within Chinese markets. However, experts believe substantial further cost-benefit analysis is needed to fully understand the economic and national security trade-offs of the proposals, along with the implications for broader U.S.-China technology competition.

At the same time, the leading chip manufacturers—SK Hynix, Micron, and Samsung—have conversely argued against federal-level intervention to address current market shortages, and against opening the door to Chinese makers. In their view, government actions affecting prices or production capacity could instead “distort” the market and risk exacerbating current challenges. The United States’ sole onshore manufacturer, Micron, currently produces less than 2 percent of its memory chip supply in the United States, though future government investments through the CHIPS Act and other targeted federal dollars for new production facilities are designed to increase that share to 10 percent in the next decade.

Q5: What is the big picture? How does the memory chip shortage figure into U.S. AI leadership and competition with China?

A5: With demand for the HBM chips fueling AI data center growth unlikely to slow anytime soon, pressure on the three U.S. and South Korean chipmakers supplying U.S. markets will continue to increase. And as it does, those manufacturers’ incentives to reserve limited chip supply for their highest-demand customers will only increase. As one indicator of future demands, by 2027, a single next-generation Nvidia processor could contain 384 GB of HBM—representing a single product consuming up to 30 percent of memory chip production capacity. For large companies such as Nvidia, hyperscalers, and cloud services providers, higher prices are not a binding constraint, and multiyear contracts reduce uncertainty about future supply. Smaller companies, universities, and even governments are less able to absorb higher costs for DRAM chips and will face higher prices and reduced access to compute capacity.

These dynamics risk sharpening disparities between the largest tech firms’ access to compute and that of smaller players in the U.S. tech ecosystem. Consumer electronics manufacturers are already experiencing increased component prices for memory chips, costs they are passing on directly to consumers. Micron, Samsung, and SK Hynix’s new factories will increase supply in several years, but will not alleviate short-term challenges.

The memory chip shortage also spotlights broader strategic challenges. The United States is the global lead in AI chip design and cloud computing, but limited domestic manufacturing capacity—today concentrated in just Micron—ultimately constrains long-term capabilities to compete with China on compute, even as data centers’ footprints are expanding across the country. Outside of Micron, two South Korean firms provide a key source of trusted supply, but the high degree of reliance on those firms exacerbates the United States’ precarious position in the semiconductor supply chain, given critical nodes across the Pacific. Looking to the future, federal government investments, including those guaranteed by the CHIPS Act, will increase U.S. resilience over time by expanding domestic production capacity and simultaneously reducing exposure to outside factors.

The memory supply shortage risks undermining the U.S. ability to generate the compute needed to compete for AI leadership vis-à-vis China. However, its biggest immediate impact will likely be on the domestic economy. Memory chip shortages that advantage the most well-off companies in the short term are increasing component costs for electronics manufacturers and their customers, all while ultimately slowing equal access to AI advancements across the economy beyond those largest companies. At the same time, the memory chip shortage shows how highly dependent U.S. AI leadership remains on access to a concentrated international supply chain with many geopolitical points of weakness.

Lauryn Williams is the deputy director and senior fellow in the Strategic Technologies Program at the Center for Strategic and International Studies (CSIS) in Washington, D.C.