Morgan Stanley recently released multiple research reports, warning of the 'approaching winter' in the memory industry, which has entered the late stage of the cycle. Among them, the oversupply of HBM (High Bandwidth Memory) may be imminent, and there is already a serious supply and demand imbalance for DRAM. Based on this, Morgan Stanley downgraded SK Hynix to 'shareholding' and 'halved' its target price, directly from 0.26 million Korean won to 0.12 million Korean won.

'Halved' the target price of SK Hynix

Morgan Stanley's research report believes that SK Hynix looks very cheap in terms of valuation and has maintained sustained competitiveness in HBM. However, the company's pricing power for DRAM is weakening, the surplus of NAND is accumulating, and the problem of HBM oversupply may soon emerge. Therefore, Morgan Stanley downgraded SK Hynix to 'shareholding', listed it as a preferred target, and 'halved' the target price from 0.26 million Korean won to 0.12 million Korean won.

According to the report, although Morgan Stanley believes that SK Hynix will still perform well for most of this year, dark clouds will gather from the fourth quarter of 2024.

First, Morgan Stanley stated that the industry's peak cycle indicators are now flashing red lights: once the current cyclical momentum of DRAM diminishes, SK Hynix may be left with a deteriorating NAND business. With the risk of normalizing production, as well as a sharp increase in capital expenditure for DRAM and HBM, Morgan Stanley believes that the risks after 2024 have been underestimated.

The report states that SK Hynix's stock appears cheap after a recent pullback, with a PB ratio of 2.0 times, and the stock price has fallen 34% from the recent high. However, Morgan Stanley expects the stock to eventually return to a PB ratio of 1.0 at the cyclical low, consistent with its historical low multiple. This implies a 51% downside valuation from the current level.

Morgan Stanley stated that the stock will undergo an adjustment of three to four quarters, in line with its pattern in all previous down cycles. At this stage, figuring out the direction of the stock is more important than trying to be right in the industry.

In addition, Morgan Stanley has downgraded YMTC from "shareholding" to "neutral", and maintains South Asia Technology, Powerchip Electronics, and Deca Microelectronics at "shareholding". Morgan Stanley has lowered its view on the South Korean technology industry to "cautious", but it still relatively "shareholding", as it is expected that quality/value will outperform the market during the downturn. $SAMSUNG EL 144 (SSNGY.US)$ is still relatively "shareholding", as it is expected that quality/value will outperform the market during the downturn.

a supply glut or performance in 2025

Morgan Stanley now sees the biggest concern in the HBM field may be the supply glut, as the market becomes more diversified and investment in the field of artificial intelligence peaks. By 2025, the current "solid" supply in the supply chain (i.e. high-quality and sufficient products) may gradually catch up with or even exceed the currently overestimated demand.

However, it is worth noting that overall, Morgan Stanley still holds a relatively positive attitude towards HBM. Morgan Stanley indicated that , $Micron Technology (MU.US)$ companies are increasing HBM's production capacity. Although profit margins may decrease, this will continue to drive strong profit margins in the AI frenzy environment, where customers are paying in advance, unaffected by the weak DRAM market conditions.

Morgan Stanley believes that in this situation, every DRAM company has the incentive to produce according to the best possible scenario for HBM output. Converting 15% of global wafer capacity to HBM requires only a small amount of capital investment (less than 10% of the wafer fabrication equipment for DRAM by 2024), but if everyone succeeds, it would be too much.

Furthermore, if there is an excess of HBM capacity, it is also a relatively easy problem to solve: use those wafers to manufacture DDR5 and idle a small portion of the backend equipment. In June, Morgan Stanley suggested that HBM3 is in a state of oversupply with a large accumulation of inventory, but this is basically not important because customers basically pay in advance, and demand is rapidly shifting to HBM3e, which may lead to shortages again.

Morgan Stanley believes this is a stable profit pool, not because HBM is a scarce commodity, but because it is more of a design-driven business that will bring a meaningful premium over standard DRAM.

The research report believes that high-bandwidth memory may become an important source of profit for Micron, in line with the market's general view of HBM. Micron has high-quality HBM3e products, and although there were some early production issues a few quarters ago, the company is now believed to be successfully increasing production capacity.

DRAM and other supply imbalances are difficult to resolve, and the industry turning point has appeared.

Morgan Stanley believes that the industry will not always be in a boom, and the cold winter will always come. Currently, although memory prices are still rising, the growth rate is approaching its peak as supply catches up with demand. The bank has adjusted the memory cycle indicator from "late cycle" to "peak cycle" for the first time since 2021, and Morgan Stanley expects industry profit growth expectations to peak and reverse over the next few quarters, with a 30% contraction in P/B ratio, increasing the likelihood of investors readjusting their positions.

According to the research report, Morgan Stanley believes that the memory industry has completed a full cycle, and the cyclical situation of DRAM may reach a plateau in the fourth quarter of 2024. Although AI demand remains relatively strong, the traditional end market has deteriorated or remained soft in recent weeks, leading to lower price expectations. Initial signs indicate that after enduring long-term third-quarter negotiations that led to further price hikes, the pricing environment in the fourth quarter will be more challenging, and there will be a trend reversal in 2025.

Morgan Stanley believes that a sell signal has now appeared, indicating that the next cyclical downturn will begin in 2025, with DRAM continuing to be in oversupply until 2026, exacerbated by inventory accumulation.

However, compared to DRAM, Morgan Stanley is slightly optimistic about NAND, believing that NAND's capacity expenditure in 2025 will still be relatively healthy. However, the bank also points out that due to similar exposure to end markets and overlapping customer bases, the cyclicality of NAND and DRAM is often somewhat synchronized, which makes the bank take a cautious attitude towards the future price trend of NAND.

Meanwhile, recent growth indicators have declined, and demand has been suppressed by weaker traditional technology. In addition, valuation is attractive in this downward process, and buying at a low point no longer makes sense. Morgan Stanley predicts that there will be better entry opportunities in the future.

Editor/Somer