Global storage stocks came under heavy pressure last week as traders reacted to three developments that appeared, at first glance, to threaten the memory-chip upcycle: Meta’s plan to offer excess computing capacity to outside customers, Apple’s testing of DRAM chips from China’s CXMT, and South Korea’s 800 trillion-won semiconductor expansion program.
A recent report from a U.S. bank argued that the selloff was driven more by fear than by a real deterioration in industry fundamentals. The report said the market may have overinterpreted the news and that each issue is either limited in scope, too far in the future to affect near-term supply, or more consistent with commercial bargaining than with a structural change in demand.
The core picture for memory remains firm, according to the report. Cloud capital spending is still rising, DRAM and NAND prices remain elevated, and major producers continue to show discipline in capacity expansion. Those factors suggest that supply and demand remain tight, even after a strong rally in storage-related shares since the start of 2026.
Still, the report also warned that the easy part of the trade may be over. With memory prices and equity valuations already at high levels, the sector is likely moving into a more selective phase. Share prices may increasingly depend on whether companies can convert strong pricing into reported profits, rather than simply benefiting from broad optimism around the cycle.
Memory selloff seen as overreaction
The latest decline in global storage stocks reflected a familiar pattern in cyclical technology markets: traders moved quickly to price in potential future risks before those risks had shown up in current data.
Meta’s disclosure that it may offer unused data-center computing capacity to external customers raised concerns that the social media and artificial intelligence group had overbuilt infrastructure or might reduce future hardware orders. Apple’s testing of CXMT memory chips was seen by some as a possible threat to established DRAM suppliers. South Korea’s massive semiconductor investment plan added another worry: that future capacity could eventually flood the market and weaken prices.
The U.S. bank report took a different view. It described Meta’s move as a monetization effort rather than a signal that AI infrastructure demand has collapsed. It viewed CXMT’s possible role in Apple’s supply chain as constrained in the short term by technology, compliance, and intellectual-property issues. It also said South Korea’s planned semiconductor expansion is spread over more than a decade, with meaningful output unlikely before the early 2030s.
That timing matters. The current memory market is being shaped by orders, wafer starts, contract prices, and cloud spending decisions taking place now. Projects that may add production several years from now do not immediately solve today’s shortages in advanced DRAM, high-bandwidth memory, enterprise solid-state drives, and legacy DDR4 products.
South Korea export data point to strong demand
Trade data from South Korea, one of the world’s most important memory-chip exporters, showed continued strength in June 2026. Semiconductor exports reached $44.8 billion, up 21% from the previous month and 199% from a year earlier. That marked the sixth consecutive month of triple-digit yearly growth.
The export figures are significant because South Korea is home to major global memory producers, and its shipment data often provides an early read on the health of the industry. The scale of the rebound suggests that demand remains broad enough to absorb higher prices, particularly from cloud computing, AI servers, enterprise storage, and data-center upgrades.
DRAM pricing has been especially strong. The report cited projections for DRAM prices to rise 53% quarter-on-quarter in the second quarter of 2026, followed by a 17% increase in the third quarter and a further 7% gain in the fourth quarter. Spot and contract levels for key products have moved beyond the highs of previous cycles. The report said 16 Gb DDR5 chips were trading near $47, while DDR4 chips were near $75.
DDR4 strength is notable because it is an older memory standard. In normal cycles, older-generation products tend to soften as production shifts to newer technology. This time, however, output reductions and limited capacity allocation have kept DDR4 supply tight, even as demand remains present in industrial systems, servers, networking equipment, embedded devices, and parts of the PC market.
NAND prices remain elevated
NAND flash memory has also recovered sharply from the lows reached during the prior downturn. The report said 512 Gb NAND wafer prices were up more than 50% year-to-date and roughly eight times higher than their February 2025 low.
Contract prices were hovering around $25, about ten times higher than early 2025 levels. Consumer storage products have followed the same direction. Prices for 512 GB consumer SSDs nearly doubled to $137.5 by June 2026.
Although the report said NAND prices appear to be stabilizing, it stressed that stabilization is taking place at high levels. That distinction is important. A market can stop accelerating while still remaining profitable for producers if pricing holds far above cost levels and supply growth stays limited.
NAND has historically been more volatile than DRAM because supply additions can be large and demand is spread across consumer electronics, enterprise storage, smartphones, PCs, and data centers. The current cycle, however, has been supported by stronger enterprise SSD demand, particularly as cloud operators expand AI-related infrastructure and as data-heavy workloads require faster storage.
Meta’s capacity plan does not imply collapsing demand
Meta’s plan to make some data-center capacity available to outside users became one of the triggers for last week’s stock decline. The concern was straightforward: if Meta has enough excess computing power to sell, perhaps it has ordered too much infrastructure, and perhaps future demand for servers, memory, and storage will slow.
The U.S. bank report pushed back on that interpretation. It characterized the plan as a way to improve asset utilization and generate revenue from capacity that may not be needed at every moment internally. Large data centers are rarely used at perfectly steady levels, and cloud operators often seek ways to balance workloads, rent unused capacity, or commercialize infrastructure.
More importantly, upstream suppliers reportedly continue to receive orders for high-bandwidth memory, LPDDR5, and enterprise SSDs. Those components are closely tied to AI servers, accelerators, and data-center systems. Continued ordering activity suggests that Meta’s external capacity plan does not amount to a broad pullback in hardware demand.
The difference between temporary unused capacity and a reduction in new infrastructure spending is central to the debate. AI workloads remain uneven, and deployment schedules can shift, but the broader buildout of compute clusters, networking, memory, and storage remains capital-intensive.
Apple’s CXMT testing viewed as limited near-term threat
The second concern came from reports that Apple has been testing DRAM chips from CXMT, a Chinese memory producer. Apple’s supply-chain decisions are closely watched because the company buys large volumes of components and often influences pricing negotiations across the electronics industry.
The U.S. bank report said CXMT’s ability to become a major Apple supplier in the near term is likely limited. The barriers include technical requirements, qualification standards, regulatory and compliance issues, and intellectual-property dependencies. Apple’s products require strict performance, reliability, and power-efficiency standards, especially for memory used across iPhones, Macs, iPads, and other devices.
The report suggested that Apple may be using CXMT’s potential participation as leverage in price negotiations with existing suppliers, rather than preparing for a rapid replacement of incumbent memory providers.
That interpretation fits a common pattern in hardware procurement. Large buyers often test alternative suppliers to improve bargaining power, diversify supply chains, or prepare for future options. Testing does not automatically mean immediate high-volume adoption.
For established DRAM suppliers, the longer-term rise of Chinese memory capacity remains a strategic issue. But in the near term, the report said the threat should not be overstated, particularly while advanced memory markets remain tight and qualification hurdles remain high.
South Korea capacity plan is a long-term issue
The third concern involved South Korea’s 800 trillion-won semiconductor expansion plan. The headline number is large and naturally raised fears of future oversupply.
However, the report said the schedule matters more than the headline figure. The projects are expected to unfold over more than a decade, and the main production impact is not expected before the early 2030s. That means the plan is unlikely to change supply conditions in 2026 or 2027.
Memory cycles can turn quickly when producers expand aggressively. But the current cycle has been marked by greater capital spending discipline, restrained wafer growth, and a focus on higher-value products such as HBM and advanced server DRAM. If producers maintain that discipline, the market could remain tight even as older capacity is repurposed or retired.
The South Korean plan may have important implications for global semiconductor leadership, supply-chain security, and long-term competition. But for near-term pricing, the report argued that it should not be treated as an immediate source of supply.
Japan checks show pricing strength
Industry checks in Japan also pointed to firm memory pricing in the second quarter. The report said DRAM and NAND average selling prices are expected to continue rising in the third and fourth quarters of 2026.
The same checks suggested that shortages could continue through 2027 if manufacturers keep capital expenditures and wafer growth under control. That would mark a major shift from earlier cycles, when large supply responses often ended up weakening prices quickly after a period of strong profitability.
The current upturn differs from previous memory recoveries because demand is less dependent on smartphone and PC restocking. Those markets still matter, but they are no longer the only drivers. AI servers, high-bandwidth memory, enterprise SSDs, and data-center DRAM have become central to the cycle.
This shift changes the structure of demand. AI infrastructure requires large amounts of specialized memory and storage. High-bandwidth memory is particularly difficult to produce and often competes for capacity with other advanced DRAM products. As a result, shortages in one part of the market can spill into others.
Samsung results highlight profit recovery
Samsung’s preliminary April-June 2026 results added another sign of the memory sector’s recovery. The company reported consolidated revenue of 171 trillion won and operating profit of 89.4 trillion won, representing year-on-year increases of 129.3% and 1,810.3%, respectively.
Full segment details had not yet been released, but the report said DRAM and NAND profitability likely played a major role in the earnings rebound. Samsung is one of the world’s largest memory producers, so its results are closely watched as a gauge of pricing power and demand strength across the industry.
The earnings figures also underline why traders have already priced in a strong recovery. When profits rebound at such a sharp pace, market expectations can rise quickly. That creates a higher bar for future performance.
In the next stage of the cycle, strong pricing alone may not be enough to support valuations. Companies may need to show that higher prices are translating into sustainable margins, stronger cash flow, and disciplined capacity planning.
Cloud spending remains the main support
The U.S. bank report estimated that Amazon, Microsoft, Alphabet, and Meta will spend about $700 billion on capital projects in 2026, nearly 80% more than in 2025. Annual spending could approach $1 trillion by 2028 if current plans continue.
That level of spending remains one of the strongest supports for the memory industry. Cloud revenue is projected to grow 35% to 40% annually, driving demand for HBM, enterprise SSDs, and data-center DRAM.
AI has changed the scale of cloud infrastructure requirements. Large language models, recommendation engines, image generation systems, enterprise AI tools, and data analytics platforms all require heavy computing capacity. Those systems rely not only on advanced processors but also on large amounts of fast memory and storage.
This creates a more durable demand base than a simple inventory restocking cycle. In a traditional consumer-electronics recovery, demand can fade once phone and PC makers rebuild inventories. In the current cycle, cloud operators are building long-lived infrastructure to support services that are expected to grow for years.
Market focus shifts to profits
The report concluded that the memory upcycle remains intact, but the market’s behavior is changing. Broad gains across the sector may give way to more selective moves based on actual financial performance.
Future stock performance is likely to depend on whether companies can maintain pricing strong enough to lift earnings, deliver profit growth that justifies prior valuation gains, and avoid the kind of capital spending surge that has ended past cycles.
That does not mean the industry has turned down. Rather, it suggests the cycle has matured. Early in a recovery, almost all producers can benefit from improving sentiment and rising prices. Later, traders tend to distinguish more carefully between companies with stronger product mixes, better cost control, deeper exposure to AI demand, and more disciplined expansion plans.
Digital assets show a similar pattern
A similar dynamic has recently appeared in digital asset markets, where headline-driven volatility has obscured a more complex supply-demand backdrop.
In June, the largest digital asset fell below $60,000 for the first time since late 2024. The decline followed roughly $4.5 billion in net outflows from U.S.-based spot market products, marking the weakest monthly flow reading on record for those vehicles. The selling pressure pushed sentiment sharply lower, with one widely followed fear gauge reaching 11 on July 1, a level associated with extreme fear.
Yet the move appeared more consistent with cyclical profit-taking than with a full exit from the asset class. Many positions had been built at lower prices earlier in 2026, giving traders an incentive to lock in gains after a strong advance. By the first week of July, flows into those products had turned positive again, and the largest digital asset recovered to trade near $64,000 by July 7.
Network data also showed a split between price action and usage. Ethereum’s on-chain transfer count reached record levels in the second quarter of 2026, with the seven-day moving average exceeding 1.3 million transactions, even as its market value fell 25.2% during the quarter.
Secondary scaling networks continued to take on a larger role. By late 2025, they were already handling about 95% of Ethereum’s total transaction throughput. That shift has pushed traders to focus less on broad narratives and more on measurable usage, fees, throughput, and revenue at the protocol level.
The digital asset market is also becoming more institutionally driven. A January 2026 survey of 351 institutional decision-makers found that nearly three-quarters planned to increase allocations, with greater emphasis on risk management, governance, and regulated access.
As with memory stocks, the next phase may be less about broad market appreciation and more about differentiation. Projects and companies with visible demand, sustainable revenue, and disciplined growth may be rewarded, while weaker names may struggle even if the overall cycle remains favorable.
For storage stocks, the main conclusion from the bank’s data is clear: the recent selloff does not yet point to a reversal in fundamentals. Cloud spending is still rising, DRAM and NAND prices remain high, and meaningful new capacity is years away. What has changed is the stage of the market. The cycle is moving from broad enthusiasm to performance-driven repricing, where earnings delivery and capital discipline matter more than headlines.
Want deeper macro context behind this memory‑chip rally? Explore market structure and liquidity trends in this analysis today.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
