Investment bank Guotai Haitong Securities has released a research report stating that the memory industry has shifted from being a beneficiary of AI computing power investments to a critical bottleneck constraining AI infrastructure development. The structural shortage driven by AI demand spans the entire supply chain, with the supply-demand gap continuing to widen.
The market's focus is expected to shift from single-quarter price and profit elasticity to the multi-year sustainability of high profitability, moving from cyclical trading based on price fluctuation expectations to a re-pricing based on the certainty of high earnings. This should drive upward movement in valuation multiples.
On the pricing front, the slowdown in contract price increases for the third quarter of 2026 is not a signal of a cyclical peak, and the upward price trend is expected to continue through 2027. Long-term agreements are not limiting the upside for price increases but rather making the price trend more predictable, exchanging short-term price elasticity for long-term visibility on earnings and cash flow.
On the supply side, the bank believes market discipline regarding supply is still underestimated. Most of the manufacturers' new capital expenditure is directed toward new factory construction and cleanroom builds. Expansion plans are being implemented in a phased manner, with production timelines remaining adjustable based on downstream demand. Actual new capacity additions in 2026 and 2027 are expected to be limited.
On the structural side, the industry has already fully priced in weak demand from consumer-facing end markets, while demand from the core data center AI theme remains robust. Under flexible capacity management, the marginal impact of traditional consumer electronics on the industry's supply-demand balance is declining and will not alter the trend of supply-demand imbalance and gradual price increases.
Key viewpoints from Guotai Haitong Securities include:
It is estimated that general-purpose DRAM contract prices will rise by 13-18% in the third quarter of 2026, and NAND contract prices will rise by 10-15%. According to TrendForce, general-purpose DRAM contract prices rose by 58-63% in the second quarter of 2026, and NAND contract prices rose by 55-60%. The forecast for the third quarter of 2026 is for a 13-18% increase in DRAM and a 10-15% increase in NAND.
In the spot market, prices for most DDR3 to DDR5 specifications continue to rise, while spot prices for some 16Gb DDR4 parts have declined. On the NAND side, TLC NAND spot prices fell by a narrower 1.5% to 1.9% month-on-month in July, QLC NAND spot prices were flat month-on-month, and MLC NAND spot prices surged 11.1% to 11.4% month-on-month. The exit or output reduction by manufacturers has led to significant price increases for MLC NAND.
South Korea's DRAM and NAND export value in June grew more than threefold year-on-year. In May, South Korea's semiconductor production, shipments, and inventory levels changed by +1.46%, -3.56%, and -7.28% year-on-year, respectively. In June, South Korea's memory export value reached $39.71 billion, a 280.3% year-on-year increase and a 23.6% month-on-month increase. DRAM exports were $21.85 billion, up 385.2% year-on-year, while NAND exports were $2.49 billion, up 301.1% year-on-year.
In May, Korean semiconductor production, shipments, and inventory levels changed by +1.46%, -3.56%, and -7.28% year-on-year, respectively. The narrowing year-on-year growth in production and a slight decline in shipments are mainly attributed to higher product value per unit.
Major memory supply chain companies in Taiwan continued to post strong revenue growth in June. Major memory manufacturers, controller chip makers, and brand/module companies in Taiwan saw their June revenues surge by 325.6%, 296.0%, and 312.0% year-on-year, respectively.
During its Q2 earnings call, Nanya Technology indicated that structural changes in the industry are widening the demand gap and expects the memory supply shortage to persist. The proportion of long-term contracts currently stands at 50%.
Risk factors identified include: the risk of technological development falling short of expectations; the risk of overheating in AI investment; risks from macroeconomic volatility; geopolitical risks; and supply chain risks.
Comments