Morgan Stanley has raised its third-quarter 2026 memory price forecasts sharply, but the bank also warned that the rally in memory-related stocks may be approaching a near-term ceiling after a powerful run driven by artificial intelligence demand, tight supply, and aggressive cloud infrastructure spending.
The bank’s latest report said pricing expectations, earnings forecasts, and portfolio positioning are already elevated across much of the memory sector. That combination, Morgan Stanley said, increases the risk of short-term pullbacks even though the longer-term demand backdrop remains supported by AI data centers, high-bandwidth memory, and expanding server workloads.
The most notable upgrade came in PC DRAM, where Morgan Stanley lifted its projected average selling price increase for the third quarter to 15% to 20%, compared with its previous forecast of 3% to 8%. Enterprise SSD pricing was also revised higher, with the bank now expecting an 18% to 23% increase. Server DRAM, GDDR6, and GDDR7 price assumptions were raised across similar ranges, reflecting continued tightness in supply and stronger-than-expected orders from AI-related customers.
The revised outlook follows a July 3 market assessment that already pointed to another strong quarter for memory pricing. That assessment projected DRAM contract prices to rise 13% to 18% quarter on quarter, while NAND Flash prices were expected to increase 10% to 15%. The latest Morgan Stanley update suggests the bank now sees the pricing cycle as stronger than previously modeled, particularly in DRAM categories linked to servers, graphics, and enterprise demand.
Still, the central message of the report was not simply bullish. Morgan Stanley cautioned that much of the good news may already be reflected in share prices and earnings expectations. For traders, that means the immediate question is shifting from whether memory prices are rising to whether the pace of improvement can still surprise the market.
Earnings expectations are already stretched
Morgan Stanley’s analysis showed DRAM earnings revision breadth approaching 89%, close to its historical peak. That means most analysts covering the sector have already upgraded their profit forecasts, leaving less room for additional positive revisions to drive prices higher in the near term.
This is an important signal in a cyclical industry. Memory stocks often move ahead of reported earnings, rising as traders anticipate price increases and margin recovery. Once forecasts become broadly upgraded, the market can become more sensitive to any sign that the rate of improvement is slowing.
The bank said valuations for memory stocks may now reflect several months of optimism tied to AI infrastructure spending and demand for high-bandwidth memory. That does not necessarily mean the cycle is ending. But it does suggest the easy phase of the rally may have passed, especially after a period in which traders aggressively rewarded companies with exposure to DRAM, advanced packaging, and AI server supply chains.
Morgan Stanley continues to favor DRAM and core memory producers because demand visibility is clearer and supply constraints remain tighter. It is less constructive on NAND and module assemblers, where pricing power and profitability appear less robust. The bank’s model projects corporate earnings growth of 35% to 40% by 2027, assuming AI-related orders continue at their current pace.
Spot prices continue to show strength
Market data shows that DRAM spot prices have risen sharply since early 2025. DDR5 16Gb prices are now near $47, while MLC 64Gb NAND prices are around $31.10. Although contract prices typically lag spot prices, both categories continue to trend upward.
The strength is being supported by demand from AI servers, cloud data centers, high-performance computing, and enterprise storage. AI workloads require large amounts of memory bandwidth and storage capacity, and cloud providers have been racing to build out the infrastructure needed to train and run large models.
That demand has changed the structure of the memory market. In previous cycles, consumer electronics such as smartphones and PCs often played a larger role in determining direction. In the current cycle, the center of gravity has moved toward servers, accelerators, and data-center infrastructure. This has made high-end DRAM and high-bandwidth memory especially important for pricing and profitability.
Even so, there are signs that the blistering pace of price increases is beginning to moderate. Third-quarter DRAM contract prices are still expected to rise by a strong 13% to 18%, but that would mark a clear slowdown from the roughly 60% increases seen in the second quarter. Consumer electronics manufacturers are increasingly resisting higher input costs, adding a point of friction to the broader pricing trend.
This tension is now one of the defining features of the market. Supply remains tight, AI demand remains strong, and producers have pricing power. But some customers are beginning to push back, especially in segments where end-demand is not as strong as in AI infrastructure.
South Korea export data highlights the boom
Fresh economic data from South Korea underscores the intensity of the semiconductor upcycle. Exports in June surged 70.9% from a year earlier to a record $102.25 billion, placing South Korea among the small group of economies, including Germany, China, and the United States, to cross the $100 billion monthly export threshold.
The record performance was overwhelmingly driven by semiconductors. Chip shipments jumped 199.5% from the previous year to an unprecedented $44.82 billion for the month. The figures show how deeply the AI infrastructure boom is feeding into national export performance and broader industrial activity.
For South Korea, the memory cycle is especially important because Samsung Electronics and SK Hynix are two of the world’s most important suppliers of DRAM and high-bandwidth memory. Their results are not only company-specific events but also indicators for the global technology supply chain.
The export surge gives strong macroeconomic confirmation that semiconductor demand is real, broad, and still accelerating in some areas. However, markets are forward-looking. Traders are now trying to determine whether this historic strength represents the middle of a multi-year expansion or a point at which expectations have become too aggressive.
Cloud spending remains the main driver
The strongest pillar supporting the memory cycle is the extraordinary capital expenditure planned by the largest U.S. cloud providers. Amazon, Google, Meta, and Microsoft have collectively raised their planned 2026 capital spending to more than $725 billion, up 77% from 2025.
That figure represents one of the most concentrated infrastructure spending cycles in history. The money is being directed toward data centers, networking, power systems, AI accelerators, memory, and storage. Because AI models require enormous computing capacity, cloud providers are competing to secure enough infrastructure to meet enterprise demand and support their own AI products.
For memory suppliers, this spending wave has been transformative. AI servers require far more advanced memory content than traditional servers. High-bandwidth memory is especially critical because it sits close to AI accelerators and helps process massive datasets at high speed. Server DRAM and enterprise SSDs also benefit as data-center workloads expand.
But the scale of spending has also brought a new risk into focus. Some analysts are questioning whether large-scale AI infrastructure spending will generate sufficient utilization and returns. If cloud providers begin to focus more heavily on efficiency, open-source models, or better use of existing compute capacity, the pace of new orders could moderate.
There are also market rumors that some large players may already be selling idle compute capacity. If confirmed, that would raise questions about whether parts of the AI infrastructure buildout are running ahead of actual demand. For now, the evidence remains mixed, and the upcoming earnings season from hyperscale cloud companies is expected to be a key test.
UBS offers a more aggressive view
Morgan Stanley’s cautious tone contrasts with a more bullish assessment from UBS. Analysts at UBS have raised their own pricing forecasts and now expect DDR contract prices to rise 32% quarter on quarter in the third quarter.
UBS argues that the memory industry will remain fundamentally undersupplied until at least the middle of 2028. Its model projects memory industry revenue reaching nearly $1 trillion in 2026 before climbing to $1.76 trillion in 2027. Under that view, the recent 17% pullback in related equity prices is not the start of a downturn but a temporary pause within a powerful multi-year cycle.
The difference between the Morgan Stanley and UBS views does not necessarily reflect disagreement about near-term demand. Both see strong pricing and tight supply. The divide is more about market positioning and durability. Morgan Stanley is warning that expectations may already be high enough to limit short-term upside, while UBS is emphasizing the possibility that supply shortages could last much longer than the market currently assumes.
For traders, that distinction matters. A strong fundamental cycle can still experience sharp corrections if expectations become too stretched. Memory stocks are known for volatility, especially when capital flows crowd into the same AI-related theme.
Volatility remains part of the cycle
Over the past two years, DRAM-linked equities have experienced three major corrections of roughly 25%, 25%, and 35%. Those drawdowns did not end the broader AI-driven cycle. Instead, they showed how volatile the sector can become when enthusiasm, earnings momentum, and positioning all rise together.
Despite those corrections, memory-related stocks have climbed from their 2022 lows to new highs. The broader direction has remained upward, supported by tightening supply and a powerful recovery in profitability. But the size of the pullbacks serves as a reminder that even strong cycles do not move in a straight line.
Morgan Stanley’s warning fits this pattern. The bank is not calling for a collapse in demand. It is saying that the combination of high expectations, elevated valuations, and crowded positioning could make the sector vulnerable to a near-term reset.
Such a reset could come from several catalysts: cautious guidance from Samsung or SK Hynix, signs of customer resistance to further price increases, weaker commentary from cloud providers, or evidence that supply is expanding faster than expected. Any one of these could increase price volatility in the sector.
Samsung and SK Hynix face key tests
Upcoming financial results from major suppliers are expected to provide the next major test of the pricing trend. Samsung has projected second-quarter 2026 revenue of about 171 trillion won and operating profit of roughly 89.4 trillion won. Market attention will focus on the strength of its memory and AI-related segments, as well as signs that profitability in traditional storage products is improving.
Samsung’s commentary will be especially important because the company has exposure across DRAM, NAND, foundry, smartphones, and consumer electronics. Traders will be watching whether management confirms continued strength in AI memory demand and whether weaker areas of the business are beginning to stabilize.
SK Hynix is scheduled to release results on July 29. Its update may carry even greater weight for the memory cycle because the company has been a key beneficiary of high-bandwidth memory demand. Markets will be closely watching commentary on third-quarter product pricing, long-term supply commitments, and capital spending plans.
Any sign that demand is cooling, customer orders are being delayed, or supply expansion is accelerating faster than expected could put pressure on the sector. On the other hand, strong guidance and reaffirmed demand from AI customers would support the view that the recent pullback is only a consolidation.
Supply expansion may eventually ease tightness
In response to sustained demand, major manufacturers are accelerating long-term expansion plans. Samsung and SK Hynix are part of a broader South Korean government-backed initiative to devote more than 800 trillion won to new advanced chip-making facilities.
The goal is to strengthen the country’s leadership in semiconductors and expand capacity in advanced memory and related technologies. Over time, those projects could help ease the supply constraints that have contributed to the current price surge.
However, new semiconductor capacity takes time to build, qualify, and ramp. That means tighter supply conditions may persist in the near term even as producers plan for future expansion. This delay between spending decisions and actual production is one reason memory cycles can become so powerful.
For now, producers appear disciplined. After the downturn that preceded the current rally, memory companies reduced output, delayed expansion, and focused on profitability. That discipline helped tighten supply just as AI demand accelerated. The key question is whether that discipline holds as profits rise and governments encourage domestic semiconductor expansion.
The market is asking a different question
The memory upcycle remains underpinned by AI data-center expansion, rising high-bandwidth memory demand, and tight supply. But the market narrative is changing. Earlier in the cycle, the main question was how much prices could rise. Now, the more important question is how long the increase can continue.
That shift is critical. When a cycle is young, positive surprises often come from stronger pricing and earnings upgrades. When a cycle matures, traders begin to focus more heavily on duration, sustainability, and the risk of overcapacity.
The next few weeks may help define that transition. Guidance from Samsung and SK Hynix will show whether memory producers still see accelerating demand into the third quarter. Earnings from the largest cloud providers will reveal whether planned AI infrastructure spending remains intact. Export data, spot prices, and contract negotiations will provide additional signals on whether pricing momentum is holding or moderating.
For now, the evidence points to a still-strong memory market with growing short-term risks. AI demand is real, cloud spending is enormous, and supply remains constrained. But expectations are also high, and high expectations leave less room for disappointment.
Morgan Stanley’s message is therefore nuanced: the structural case for DRAM and core memory producers remains attractive, but the near-term risk-reward has become less one-sided. Traders may need to distinguish between a healthy consolidation after a major rally and the early signs of a broader reassessment of the cycle’s longevity. The answer will likely come from company guidance, cloud capital spending plans, and whether memory prices can keep rising without triggering stronger resistance from customers.
As AI and equities intertwine, sharpen your edge by exploring advanced market tools with technical analysis in stocks vs crypto 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.
