AI Stocks Face Broad Selling: What Does a 1.6x Annualized P/E Mean for Memory Stocks?

The memory chip industry is an important part of the semiconductor supply chain. It mainly includes DRAM, NAND Flash, and high-bandwidth memory (HBM) designed for AI accelerated computing. As AI data center construction continues to expand, memory affects not only server capacity and performance, but also plays a key role in the efficiency of AI accelerator compute.

Recently, a market view from Serenity suggested that the AI industry chain is facing broad stock sell-offs. For some memory stocks, the price-to-earnings ratio based on annualized recent profits can be as low as 1.6x. This view raises a question worth discussing: if a company’s earnings are growing rapidly while its stock price keeps falling, does an extremely low P/E ratio mean the stock is already undervalued?

For investors following Micron Technology (MU), Seagate Technology (STX), Western Digital (WDC), and other memory-related companies, the key is to distinguish between current earnings levels and sustainable earnings capacity. This article analyzes the opportunities and risks that an extremely low valuation may reflect, focusing on annualized P/E, the memory industry cycle, and changes in AI demand.

Serenity’s 1.6x P/E: which company is it referring to?

In the market commentary circulated on November 5, Serenity (@aleabitoreddit) used Taiwan memory chip company ESMT (G-Super Microelectronics; 晶豪科技) as an example, arguing that the market may be underestimating the earnings support stemming from a contraction in traditional memory supply.

The related report cites the following data.

ChainCatcher+1

These numbers require extra caution. The report expresses the financial data in a unified currency of U.S. dollars, but ESMT is a Taiwan-listed company. Its original financial statements are usually reported in new Taiwan dollars. You must verify the currency and the conversion methodology to confirm whether the U.S. dollar figures above are accurate.

The logic behind a 1.6x P/E calculation is as follows: assume a market cap of $2 billion and quarterly net profit of $330 million. Multiply the quarterly profit by four to get an annualized net profit of $1.32 billion. Then divide market cap by annualized net profit to get about 1.52x. After rounding the data and accounting for market-cap fluctuations, the valuation could come out to roughly 1.6x.

However, this calculation hides an important assumption: the company can maintain the current quarter’s earnings level for the next four quarters. If Q3 profits benefited from a rapid rise in memory prices, then the annualized result could be far higher than the company’s earnings power under a normal cycle. Therefore, a 1.6x P/E should not be taken directly as a forward P/E forecast for the next 12 months, and it cannot prove that the stock is undervalued.

Why could sell-offs across the AI industry chain affect memory chip valuation?

The AI industry chain spans many areas, including GPUs, servers, memory, networking equipment, optical communications, and data center infrastructure. Even though these industries can all benefit from AI investment, their revenue sources, cost structures, and profitability cycles differ.

When market risk appetite falls or institutional investors reduce leverage, investors may sell stocks across multiple AI-related sectors at the same time, without fully distinguishing based on each company’s fundamentals.

Serenity describes the current market as indiscriminate selling. It believes some companies’ stock prices have already fallen back to levels before major orders were secured and profit expectations were realized. This is its interpretation of market trading behavior, not proof that all AI stocks have already seen fundamentals and valuations fully decouple.

For memory chip companies, a drop in stock prices may include two types of factors at once: one is short-term selling pressure from overall deleveraging in the market, and the other is investors’ concerns that the cycle of rising memory prices cannot be sustained long term.

Only by separating these two factors can you judge whether the undervaluation is likely to persist.

Why could DDR2 and DDR3 supply contraction support ESMT’s earnings?

ESMT’s business includes traditional memory products such as DDR2 and DDR3. Unlike HBM, which is built for the latest AI accelerators, these products are used more in industrial equipment, embedded systems, and some mature consumer electronics.

Serenity argues that as major memory manufacturers like Samsung and Micron gradually reduce capacity for traditional products, supply contraction could emerge in the relevant market. Its view is that even if end-consumer demand declines somewhat, as long as supply decreases faster, DDR2 and DDR3 prices may still be supported.

This logic differs from the commonly discussed idea that AI-driven memory demand will grow. HBM growth is mainly driven by AI compute demand. Meanwhile, price increases in traditional DDR products could also stem from production line exits and changes in supply concentration.

That said, supply contraction does not guarantee that prices will keep rising. If customers shift to substitute products, inventory demand weakens, or existing manufacturers ramp up production again, traditional memory prices can still fall. Therefore, when analyzing ESMT, you need to focus on the actual average selling price, volume, inventory, and customer demand for DDR2 and DDR3—not simply equate AI data center investment growth with growth in demand for all of its business.

Why doesn’t a 1.6x annualized P/E necessarily mean the stock is cheap?

The P/E ratio reflects the relationship between a stock price and a company’s earnings. For companies with relatively stable profits, the P/E ratio helps investors compare valuation levels. But for cyclical industries such as memory chips, P/E ratios can sometimes be misleading.

When memory prices rise, a company’s profits can increase quickly, causing the P/E based on annualizing the most recent quarter’s profit to drop just as fast. However, if memory prices fall in the future, company profits could drop significantly. In that case, what looked like an extremely low P/E could rise. Take a company with a $2 billion market cap as an example:

The scenarios above assume a fixed market cap. They are not ESMT’s earnings forecast. This shows that a P/E ratio depends not only on the stock price, but also on what earnings level investors use as the denominator. For memory companies, investors need to pay more attention to earnings power across cycles, rather than simply extending a single quarter’s profits at a favorable point in the cycle into the full year ahead.

How is the market logic different for HBM, DRAM, and NAND?

Even though these products all belong to the memory supply chain, demand drivers and competitive structures differ significantly.

HBM demand is mainly tied to the deployment of AI accelerators. As high-performance GPUs demand higher memory bandwidth, HBM becomes an important component of compute systems.

Traditional DRAM may benefit as major manufacturers adjust their product mix, which reduces supply. When companies shift resources to newer products, older-spec memory modules may face temporary supply tightness.

NAND Flash has a different market logic. Its demand is influenced by enterprise SSDs, data center storage, and consumer electronics products.

Therefore, even if the broader AI industry chain experiences sell-offs, you cannot conclude that all memory companies face the same changes in earnings. Investors need to analyze product mix and the supply-demand environment separately.

Can long-term supply agreements reduce cyclical risk for memory companies?

Serenity also mentioned that companies such as Micron, Samsung, SK hynix, and SanDisk are improving the predictability of future revenue and pricing through long-term supply arrangements. Long-term supply agreements may help companies stabilize customer relationships and reduce some uncertainty caused by short-term price swings. For cloud computing firms and equipment manufacturers that need to purchase large volumes of memory continuously, locking in supply plans early also helps secure production schedules.

That said, the level of protection provided by different contracts can vary greatly. Some agreements focus mainly on purchase quantities. Some may include pricing adjustment mechanisms. Others may set minimum purchase volumes or other conditions.

Therefore, you cannot assume that a company’s future profits are locked in just because it has long-term contracts. Actual outcomes still depend on contract duration, pricing methods, fulfillment obligations, and customer credit conditions. For the memory industry, long-term agreements can improve revenue visibility, but they typically cannot fully eliminate risks arising from technological iteration, price changes, or demand cycles.

How to tell whether there’s a valuation-repair opportunity in AI memory stocks?

When memory companies’ stock prices fall and their annualized P/E ratios drop significantly, investors can analyze from two angles: earnings quality and industry supply-demand.

• First, confirm whether profit growth comes from higher unit shipments, higher product pricing, or one-time gains. If profits are mainly driven by short-term price increases, you need to assess whether future prices can hold up.

• Next, focus on cash flow. Profit growth in net terms does not necessarily mean operating cash flow rises at the same pace—especially when there are major changes in inventory, accounts receivable, and capital expenditures.

• Then, evaluate whether supply changes are likely to be sustained. If traditional memory capacity permanently exits while customers still need the relevant products, price support could last longer. Conversely, if rising prices attract new supply, earnings power could come under renewed pressure.

• Also consider valuation and the balance sheet. A higher net cash position may provide some financial buffer, but it doesn’t guarantee the stock price won’t keep falling.

How to track valuation changes in the AI memory supply chain via Gate stock?

Market adjustments across the AI industry chain not only affect GPU and data center companies, but may also spill over to memory chip firms, semiconductor manufacturers, and related infrastructure providers. For investors who want to research this space, you can use Gate stock to track relevant U.S.-listed companies, such as Micron Technology (MU), SanDisk (SNDK), Western Digital (WDC), Seagate Technology (STX), and Nvidia (NVDA). The specific tradable tickers should follow what the Gate platform actually displays.

Different companies have different business positioning within the AI industry chain. Micron Technology mainly involves memory products such as DRAM, NAND, and HBM. SanDisk and Western Digital each cover flash and hard-disk storage businesses. Seagate Technology mainly provides high-capacity data storage solutions. Nvidia, meanwhile, sits in the AI accelerated computing layer. As a result, even if these related stocks are all influenced by market sentiment at the same time, their earnings drivers and valuation logic are not identical.

For the AI memory supply chain, what truly matters is whether companies can turn rising memory prices, AI demand growth, and changes in supply structure into sustainable profits and free cash flow. Gate stock can serve as a channel to observe and participate in the relevant U.S. stock market, but any specific trades still need to factor in risks such as valuation volatility, industry cycles, and product-specific rules.

Summary

Serenity’s view that the AI industry chain is facing indiscriminate sell-offs reflects today’s disagreements in the market about the valuations of semiconductor and memory companies. Some memory firms may still deliver strong quarterly earnings, yet their stock prices can still drop due to market deleveraging, cyclical concerns, and shifts in risk appetite. Taking ESMT as an example, the ~1.6x annualized P/E can be visually striking, but its calculation depends on extrapolating recent-quarter profits across the whole year. In cyclical industries like memory chips, this valuation approach may understate the risk of future profit declines.

Meanwhile, the exit of traditional DDR2 and DDR3 capacity, HBM demand growth, and the development of long-term supply agreements can indeed change some memory companies’ supply-demand structure and earnings stability. Therefore, when assessing whether AI memory stocks are undervalued, the most important factor is not how low the current P/E is, but whether the company can continue to generate the corresponding profits over the next few quarters and translate those earnings into real cash flow.

FAQ

Is ESMT’s annualized P/E really only 1.6x?

Serenity cites a valuation of about 1.6x, but it is based on an annualized calculation of a single quarter’s profits, not a fully validated P/E after comprehensive earnings forecasting. The currency and measurement basis of the relevant financial figures also need further verification.

Why could memory stocks’ prices keep falling even if profits grow?

Because the stock price reflects the market’s expectations for future earnings. If investors believe current profits are at a cyclical high point, even strong performance in the latest financial report may not be enough to support the stock price.

Is ESMT in the same memory business as Micron Technology?

Both companies deal in memory products, but they differ in product structure, customer markets, and competitive strengths. ESMT covers traditional and special-purpose memory products, while Micron has a broader lineup across DRAM, NAND, and HBM.

Will AI demand growth definitely push up the prices of all memory chips?

Not necessarily. Different memory products have different supply-demand structures. HBM demand is highly tied to AI accelerators, while traditional DDR and NAND prices are still influenced by consumer electronics demand, capacity changes, and inventory cycles.

Does a low P/E automatically mean a stock is a good buy?

You can’t reach that conclusion directly. A low P/E may indicate lower valuation, or it may reflect market expectations for declining future profits. You need to evaluate it together with normalized-cycle earnings, cash flow, liabilities, and industry competition.