BiyaPay Research Institute Analysis: US Stocks' Black Thursday, Nasdaq Falls 5.52% — Why Did AI Memory Stocks Drop Even Harder?

AI stocks are still holding up, but share prices fell first.

On November 5 local time, the US stock market saw a round of violent selling. According to BiyaPay market data, the Nasdaq Index fell 5.52% in a single day, closing at 24,315 points; the S&P 500 Index fell 4.36%, and the Dow Jones Industrial Average fell 3.44%. Tech stocks and the semiconductor sector became the main drag, with the Philadelphia Semiconductor Index falling more than 12%.

Individual stocks performed even worse. SanDisk (SNDK) fell 22.26% in a single day, Micron Technology (MU) fell 17.46%, Western Digital (WDC) fell 17.35%; AMD fell 17.26%, and Nvidia also fell 9.26%.

The most notable aspect of this decline is not just how large the drop was, but the scope and direction of the market selloff: from chip design and storage devices to data center-related companies, multiple segments of the AI supply chain came under pressure simultaneously.

The sharp market volatility has also made investors pay more attention to capital arrangements across different markets. For users who follow both digital assets and traditional financial markets, how to bridge USDT with markets such as US stocks and Hong Kong stocks is also a question worth considering in global asset allocation. BiyaPay, a global one-stop asset allocation platform, supports multi-currency exchange and provides access to investment markets such as US stocks and Hong Kong stocks. Users can use USDT as a funding entry point, exchange it into US dollars or Hong Kong dollars within the platform, and then make investment arrangements according to the corresponding account and trading rules. This funding linkage helps reduce operational steps between different platforms and makes multi-market capital management clearer.

Previously, the market was trading on expectations of continued growth in AI demand, rising capital expenditure by tech giants, and accelerated data center expansion. Now, investors are starting to recalculate another set of questions: can these investments be implemented on schedule? Will infrastructure construction be constrained by policy and energy conditions? How much profit can massive capital expenditure ultimately translate into?

This may be the more important change behind this US stock market selloff.

Why did the Nasdaq suddenly plunge? The midterm elections were only the trigger; AI investment concerns had long existed

The US stock selloff on November 5 is closely related to changes in policy expectations after the US midterm elections, but if this decline is attributed entirely to the elections, it underestimates the pressure that had already built up within the market.

Before the election results were announced, the AI sector was already facing several controversies: tech giants continue to increase capital expenditure, but can AI commercialization revenue keep up? Data center construction requires large amounts of power and infrastructure, so is the project payback period too long? Have the high growth expectations for chip companies already been fully reflected in share prices in advance?

These questions did not emerge only on November 5; the market had simply been willing to tolerate higher valuations before.

After the US midterm election results were announced, the Democratic Party achieved important election gains, and the market began to reassess the future policy environment for AI and data center construction in the United States. Some local elections and referendums also pushed issues such as data center land use, energy consumption, water use, and community impact into the spotlight.

The market is worried that some projects may face stricter approvals, longer construction cycles, or even suspension or adjustment in the future. It should be noted that policy discussions and local-level demands for restrictions do not mean the United States has fully halted data center construction. Policy authority and project conditions differ across states and localities, and this cannot be simply understood as a full stop to data center construction.

But capital markets often trade on expectations rather than waiting for policy to actually land before reacting.

For the AI supply chain, even if only some projects are delayed, this could affect the order pace and revenue recognition of related companies in the coming quarters. Once uncertainty emerges in growth expectations previously regarded as highly certain, valuations may come under pressure first.

More importantly, before the election, the market already had problems such as crowded AI trades, elevated valuations, and doubts about returns on capital expenditure. The election results were more like a catalyst that allowed previously accumulated concerns to be released in a concentrated manner.

Why did AI storage stocks fall harder? The market is worried about demand, prices, and the profit cycle at the same time

If the overall decline in the AI supply chain reflects a decline in market risk appetite, then the sharp drop in storage stocks also involves the industry's own cyclical characteristics.

AI model training and inference require massive data processing capability. GPUs handle computing, HBM provides high-bandwidth memory support for processors, and DRAM and NAND handle data storage and read/write tasks at different levels. As the scale of AI servers and data centers expands, the importance of storage devices also increases.

This is also why storage companies have been sought after by capital for some time.

But in the storage industry, corporate profits do not keep growing simply because demand grows. The industry is also affected by factors such as supply expansion, product prices, capacity utilization, customer inventory, and product mix.

This adjustment can be understood from three levels.

First, growth in AI demand does not mean all storage products benefit simultaneously.

HBM, server DRAM, enterprise SSDs, and traditional hard drives face different customer demand, technical barriers, and supply-demand environments. Even if AI data centers continue to expand, order growth and profitability may diverge across different products.

Second, storage prices and profit margins are cyclical.

When industry supply is tight and product prices rise, storage manufacturers' profitability may improve rapidly; but if companies expand production, supply increases, or customer procurement pace changes, product prices and profit margins may come under pressure.

What the market is really worried about is whether the current high profitability can be sustained, not just how many orders remain today.

Third, the market may trade changes in future profit expectations in advance.

Stock prices reflect future expectations, not a company's current operating conditions. If investors believe that product prices, profit margins, or order growth in the coming quarters may be lower than previously expected, share prices may fall in advance even if the company's current performance remains strong.

Therefore, the sharper decline in storage stocks does not mean AI storage demand has disappeared, but that the market has begun to reassess how long this boom cycle can last.

They are all storage stocks, so how do the decline logics of Micron, SanDisk, and Western Digital differ?

The simultaneous plunge in Micron, SanDisk, and Western Digital this time makes it easy to see them as the same kind of AI storage asset. But from a business structure perspective, the three companies are not entirely the same.

Micron Technology mainly operates DRAM and NAND flash products, among which HBM is an important component of AI servers. The high-performance memory demand brought by AI data center expansion is an important reason why the market pays attention to Micron.

SanDisk mainly focuses on NAND flash and related storage products, and its business is closely related to flash memory demand, product prices, and the supply-demand cycle.

Western Digital mainly focuses on the hard drive business. As data volumes in data centers continue to increase, enterprise hard drives still have room for application, but their product structure and profit drivers differ significantly from Micron's DRAM and HBM businesses.

Therefore, the three companies cannot simply be classified as HBM suppliers, nor can it be assumed that they will show exactly the same fundamental performance due to the same change in AI demand.

This is also where further breakdown research is needed after the sector plunge: for Micron, look at the competitiveness, supply, and prices of its HBM and DRAM businesses; for SanDisk, look at NAND product prices and inventory; for Western Digital, focus on enterprise hard drive demand, product structure, and profitability.

Sector sentiment can change in sync, but company fundamentals may not.

Is data center construction really about to cool down? Industry data shows another side

Market concerns about data center project delays do not mean AI infrastructure demand has disappeared.

CBRE's North America Data Center Report for the first half of 2026 shows that in the eight major North American data center markets it tracks, capacity under construction reached 7,481 megawatts, up 24.8% from the previous reporting period; more than 80.4% of capacity under construction had already secured pre-lease commitments.

The report also shows that total supply in these major markets reached 10,903 megawatts, with the vacancy rate falling to 1.4%. This reflects that data center capacity remains tight, and demand for infrastructure from AI and large cloud service providers remains strong.

This set of data shows that data center construction has not lost its demand base because of market concerns.

But it also reveals another easily overlooked problem: strong demand does not mean all projects can be delivered on schedule.

Data center construction requires not only land and capital, but also stable power supply, supporting facilities, permits and approvals, and sufficient construction capacity. The CBRE report also points out that power availability, infrastructure delivery cycles, and opposition from local communities are becoming important factors affecting project site selection and construction.

This means the data center industry may see more obvious divergence in the future: projects that can secure power resources, successfully complete approvals, and deliver on time still have opportunities to benefit from strong demand; while projects lacking power conditions, facing approval obstacles, or with excessively high financing costs may face delays and rising costs.

So what really needs attention is not whether there is demand for data centers, but whether demand can be smoothly converted into deliverable capacity and operating revenue.

Micron's historical P/E ratio is only about 8.8x, so why did its share price still suffer a heavy blow?

In this adjustment, Micron's valuation also deserves separate discussion.

According to Micron's announced fiscal 2026 results, the company's full-year revenue reached $133.188 billion, and GAAP diluted earnings per share were $74.33. Based on the closing price of $656.39 on November 5, the historical P/E ratio is about 8.8x.

Looking at this number alone, Micron does not seem expensive. But the problem is that the storage industry is a typical cyclical industry, and current-period profit alone cannot be used to judge whether the valuation is reasonable.

When the industry is in an upcycle, product prices rise and capacity utilization improves, and corporate profits may grow rapidly. At this time, the P/E ratio calculated using high profits often appears relatively low.

But if the market expects future product prices to fall and profit margins to narrow, corporate earnings may change, and even if the current P/E ratio is not high, investors may still lower valuations in advance.

This is a phenomenon that often appears in cyclical stocks: when profits are at their best, the P/E ratio may look lowest; and what the market is really worried about is precisely whether current high profits can be sustained.

Therefore, Micron's P/E ratio of about 8.8x is calculated based on already announced historical earnings for fiscal 2026, not a forecast of future earnings, and cannot be directly used to conclude that the stock price is cheap.

More worth watching is whether the company's HBM shipments, DRAM and NAND prices, gross margins, capital expenditure, and long-term customer procurement arrangements in the coming quarters can continue to support profitability.

It is not just fundamental changes; how did concentrated capital outflow amplify the decline?

There is another factor that cannot be ignored: this decline may not only reflect investors reassessing corporate value, but also involve concentrated capital withdrawal from popular trades.

The AI supply chain had previously experienced a relatively long rally, some tech stocks had high valuations, and market trading was relatively crowded. When policy expectations, earnings concerns, and the interest rate environment change at the same time, investors may choose to reduce risk exposure.

If institutional investors adjust positions simultaneously, or if some trading strategies need to reduce leverage, sell orders may appear in concentrated fashion within a short period.

Such capital behavior amplifies price volatility, especially in stocks that had previously risen significantly and where market positioning is concentrated.

But it is necessary to distinguish that a rapid share price decline cannot directly prove that large-scale forced liquidations occurred in the market, nor can it be used to infer that certain funds have already experienced liquidity crises. Judging whether capital is undergoing passive deleveraging still requires information such as trading volume, margin data, fund holdings, and derivatives market conditions.

For ordinary investors, what matters is recognizing that short-term share price fluctuations are not necessarily entirely determined by changes in company fundamentals; market sentiment and capital structure can also significantly accelerate the pace of decline.

What to watch after the plunge? Whether AI storage stocks can stabilize depends on these indicators

After this round of sharp adjustment, rather than rushing to judge whether the AI rally is over, it is better to focus on several more valuable indicators.

First, observe tech giants' capital expenditure plans. If large cloud service providers continue to maintain AI infrastructure investment, it shows that long-term demand is still supported; if companies begin to cut budgets, delay projects, or adjust construction pace, the supply chain's growth expectations need to be reassessed.

Second, observe storage product prices and gross margins. For companies such as Micron and SanDisk, product prices, order structure, inventory, and capacity utilization often reflect industry prosperity better than single-day share price performance.

Third, pay attention to the actual delivery of data center projects. Pre-leasing and planning can only indicate potential demand; whether power can ultimately be secured, construction completed, and operations launched determines whether related investments can generate actual revenue.

Fourth, pay attention to AI commercialization and cash flow. Whether AI companies can obtain sustained revenue through model services, cloud computing, software, and enterprise applications will directly affect the market's acceptance of large-scale capital expenditure.

Fifth, observe whether capital flows back into the tech sector. If a subsequent market rebound is driven only by a few heavyweight stocks while the storage, semiconductor, and data center supply chains remain weak, it would indicate that disagreement over the AI investment logic may not yet be over.

Is the AI bull market over? The market may be entering a phase that places more emphasis on profitability

The US stock market plunge on November 5 cannot be simply understood as the AI industry peaking, nor can it be assumed that this adjustment is only short-term sentiment just because data center demand remains strong.

A more reasonable understanding is that the market is reexamining the growth quality of AI investment.

In the past, investors focused more on growth in computing power demand, data center expansion, and tech giants' capital expenditure; now, the market is beginning to ask whether projects can be implemented on time, whether companies can continue to secure orders, whether product prices and profit margins can be maintained, and whether massive investment can ultimately be converted into cash flow.

This means the AI supply chain may no longer see a market where all companies rise together, but may gradually enter a stage that places greater emphasis on profit realization, product competitiveness, and return on capital.

For storage companies such as Micron, SanDisk, and Western Digital, long-term demand is still worth tracking, but whether short-term share prices can stabilize will ultimately depend on whether orders, prices, inventory, and profits can withstand market scrutiny.

The long-term AI story may not be over, but the phase in which growth stories alone could drive share prices higher may be changing. Next, what the market really needs to test is who can turn AI demand into sustained, realizable profits.