Storage Chip Valuation Debate: Asian ADR in a High Cycle, Is it PE or PB?

Entering the fourth quarter of 2026, the Asian storage chip sector's ADR performance in the US stock market continues to be strong, but unlike the previous atmosphere of mere "chasing gains," the focus of increasing numbers of institutional investors is shifting from "how much more can it rise" to "is it expensive now." A noteworthy phenomenon is that some storage leaders have forward earnings multiples based on current earnings that are not high, but instead are in the lower end of historical ranges. For investors accustomed to the view that "low valuation equals a safe margin," is this a rare valuation trough, or a "low PE trap" typical of cyclical stocks? ## Low PE Trap: The Old Problem of Storage Cyclical Stock Valuation Storage chips are one of the most cyclical segments in the semiconductor industry, with prices driven by supply and demand shortages; they soar when supply is tight and crash severely when demand slackens. This boom-and-bust cycle causes systematic biases when observing storage stocks through a traditional PE lens: high earnings peaks often correspond to low PE points, while low earnings points correspond to high PE points. Looking back at the last two upswings in the storage cycle, around the peaks of DRAM prices, leading companies generally had valuation multiples based on current earnings at historical lows, but then a multi-year downtrend saw stock prices fall significantly. In other words, a seemingly cheap PE is precisely the least reliable signal at the top of a cycle. This is why experienced cyclical stock investors tend to focus on Price-to-Sales ratios, Enterprise Value multiples, and earnings median assumptions, rather than just the current P/E ratio. ## Why the Valuation Framework Needs Reconstruction in This Cycle If historical experience supports the warning "high earnings, low PE," then several structural changes in this cycle mean that simply applying the old framework might also be incorrect. ### HBM Changes Earnings Structure High Bandwidth Memory (HBM) adopts a model of long-term agreements with cloud vendors and AI chip manufacturers, with some orders involving prepayment arrangements, and capacity is often locked in before production starts. Compared to the price competition in spot DRAM, the revenue visibility for HBM business is significantly higher, and the fluctuation in price and utilization rate is smaller. This means the earnings median for leading companies is systematically lifted, and the volatility of earnings decreases—theoretically, this should support a valuation multiple higher than the historical average. ### More Diversified Demand Structure In the past, storage demand was highly dependent on smartphones and PCs, with a relatively thin demand curve, meaning that if the end market weakens, the entire supply chain is under pressure simultaneously. In this cycle, the pull from AI servers for HBM and large-capacity DRAM, and the pull from enterprise SSDs for inference applications, means that demand sources are clearly diversified. The more diversified the demand structure, the smaller the impact of a downturn in a single application on the overall cycle, which is the second reason supporting the upward shift of the valuation median. ### Oligopoly Discipline Strengthens on the Supply Side After multiple rounds of pruning in the storage industry, the supply of DRAM and NAND is highly concentrated among a few manufacturers. After experiencing deep losses, the major manufacturers have shown stronger discipline in capital expenditure; the blind expansion based on "trading price for volume" is clearly converging. Changes in supply-side behavior are the most critical variable for determining whether the earnings median can be sustained in this cycle, and it is a prerequisite that cannot be bypassed in valuation discussions. ## PB-ROE Framework: The Ruler More Suitable for Cyclical Leaders For cyclical industries, the PB-ROE framework better captures the essence of "stock price reflects the sustainability of return on net assets" than PE. The logic is simple: when a company's ROE consistently remains significantly above its cost of capital, its PB should be higher than its historical median; Conversely, if ROE rebounds are just fleeting, PB will inevitably revert to the mean. Therefore, the core of judging whether a valuation is reasonable is not the current PE figure, but how long high ROE can be sustained. Furthermore, dividend and buyback policies are important windows into management's confidence. When a storage leader chooses to increase dividend payout ratios or initiate large-scale buybacks during a high-cycle period, it is essentially a "vote of confidence" in the sustainability of its cash flow. These signals have more persuasive support for valuation than single-quarter financial figures, making them worth tracking for ADR investors. ## Pricing Characteristics in the ADR Market: Triple Premium Investors participating in Asian storage leaders via US ADRs also need to understand the differences in ADR pricing compared to local markets. First is the exchange rate factor;the trend of the US dollar against Asian currencies directly affects the premium on ADRs relative to local stocks. Second is the liquidity premium; if some ADR trading activity is lower than the local market, large inflows and outflows need to consider the impact cost. Third is the funding factor; changes in foreign ownership ratios and the flow of funds into related ETFs are high-frequency thermometers for observing the attitude of international pricing funds. When a clear divergence occurs between the local market and ADR trends, it often signals that capital is re-pricing cyclical expectations, which itself is a signal worth paying attention to. ## Four Signals Investors Should Track * Price Structure: The direction of the difference between spot price and contract price, and the change in the proportion of long-term contract orders in revenue, directly reflects the "quality" of earnings. * Inventory Levels: Downstream channel inventory weeks are one of the most reliable leading indicators of the cycle. Healthy inventory indicates stronger price resilience, whereas a vacuum in demand after restocking should be watched out for. * Capital Expenditure and Dividends: Expansion guidance determines the future supply-demand balance, and the strength of dividends and buybacks reflects management's true judgment of the cycle. * Technology Iteration Pace: The certification and mass production progress of the next generation of HBM products determine whether the leader can continue to joy a structural premium and whether laggards can close the gap. Of course, there is no single answer to the debate on valuation. If new capacity is released on the supply side and AI capital expenditure slows down, storage price elasticity may once again dominate stock prices, and valuations will face pressure to revert to historical averages. Conversely, if the proportion of long-term contracts and demand diversification continue to deepen, the market may gradually grant Asian storage leaders a new pricing of "cyclical weakening and growth reinforcement." For investors, rather than dwelling on "low PE means cheap" or "sell at the peak of the cycle," it is better to combine the judgment of the cyclical position with valuation discipline: anchor the reasonable range with PB-ROE, dynamically adjust with inventory, price structure, and capital expenditure signals, and then choose the tool for participation based on the ADR premium. In an industry where the cycle is king, valuation is never a static number game; it is the probability pricing of future supply and demand patterns—only by understanding the cycle can one understand valuation