AI Stocks Surge, Then Plunge: Which of 4 Sectors Is Worth Buying?

• Micron has gained 324.5% over the past year, while Seagate and Western Digital have risen more than 170%, but all three have experienced substantial corrections from their highs.
• Applied Materials and Lam Research are down approximately 47% from their highs, despite continued demand for advanced semiconductor manufacturing technology.
• Nvidia, TSMC, Amazon, and Meta have experienced much smaller corrections, suggesting investors are distinguishing among four different AI-related business models.

AI Stocks Are No Longer Moving Together

Artificial intelligence has driven extraordinary investment across the semiconductor and data center industries, but the financial benefits have not been distributed evenly among the companies supplying AI infrastructure. Semiconductor memory manufacturers, hard disk drive suppliers, wafer fabrication equipment companies, and AI computing platforms occupy different positions in the supply chain, and their stock prices increasingly reflect those differences.

Over the past year, Micron Technology (MU | MU Price Prediction) has appreciated 324.5%, while Seagate Technology (STX) and Western Digital (WDC) have gained approximately 179% and 173%, respectively. Applied Materials (AMAT) and Lam Research (LRCX) have also generated substantial gains, but their stocks have subsequently declined nearly 47% from their respective highs. Nvidia (NVDA) and Taiwan Semiconductor Manufacturing Company (TSM) meanwhile, have experienced much smaller corrections, despite being central to the same AI investment cycle.

These differences are particularly revealing because AI infrastructure requires products from all four segments. Nvidia supplies processors, TSMC manufactures advanced chips, Micron and SK hynix (SKHY) provides memory, and Seagate and Western Digital supply mass-capacity storage. Applied Materials and Lam Research manufacture equipment required to produce semiconductors. Amazon (AMZN) and Meta Platforms (META), as major purchasers of AI infrastructure, represent companies financing much of the demand flowing through the supply chain.

According to Table 1, the ten companies exhibit four identifiable business groupings, with notable similarities in both one-year appreciation and subsequent declines from their highs. The comparison also reveals that the industries generating the strongest stock appreciation are not necessarily those experiencing the smallest corrections.

The most important observation is that stock performance is clustering by business model. Micron and SK hynix have experienced extraordinary appreciation associated with memory demand and pricing, followed by corrections of approximately 28%–31%. Seagate and Western Digital have delivered almost identical one-year gains of approximately 175%, but their corrections are substantially larger, averaging 48%. Applied Materials and Lam Research form another closely aligned group, with gains of 56%–65% followed by declines of approximately 47%. Finally, Nvidia, TSMC, Amazon, and Meta have remained within 12%–17% of their highs, despite significant differences in their one-year returns.

These patterns suggest that investors are distinguishing among companies according to the economics of their businesses, their capital requirements, and their exposure to future AI infrastructure spending. Understanding why each segment appreciated, and why investors have subsequently reduced valuations, is essential to evaluating whether the corrections represent deteriorating fundamentals or investment opportunities.

Segment 1: Semiconductor Memory — Strong Earnings, but Are Profits Near a Peak?

Semiconductor memory has benefited from one of the strongest pricing environments in years. AI accelerators require high-bandwidth memory, while cloud servers and enterprise computing systems consume conventional DRAM and NAND. Expanding HBM production also absorbs manufacturing resources that could otherwise support conventional DRAM output, tightening supply across the industry. This combination has produced extraordinary earnings growth, but investors are increasingly concerned about how long favorable pricing can continue.

Micron Technology: Is a 324.5% Winner Still a Buy?

Micron’s 324.5% one-year stock appreciation reflects the powerful combination of AI memory demand, higher selling prices, and improving manufacturing profitability. The company’s HBM business benefits from increasing AI accelerator shipments, while its conventional DRAM and NAND operations benefit from tight industry supply. Because memory manufacturing carries substantial fixed costs, rising prices can produce disproportionately large increases in operating earnings.

Micron’s September 30 earnings release reported fiscal 2026 revenue of $133.19 billion, compared with $37.38 billion in the previous year. Management also projected first-quarter fiscal 2027 revenue of approximately $61.5 billion, illustrating the extraordinary momentum in the memory market.

Despite these results, Micron has declined 27.7% from its high, including approximately 20% during October according to Moomoo. The concern is not necessarily weakening current demand, but whether investors are extrapolating unusually favorable memory pricing too far into the future. Additional production from Samsung, SK hynix, and Chinese manufacturers could eventually moderate prices, while the company’s expanding capital requirements may affect future free cash flow.

Moomoo estimates that Micron trades at approximately 4.9 times expected fiscal 2027 earnings. That valuation appears attractive, but memory companies frequently trade at low earnings multiples near cyclical profit peaks. The critical question is whether HBM demand and constrained conventional memory capacity can sustain earnings longer than investors currently anticipate.

Micron looks attractive for investors who expect memory shortages to persist into 2027, but its exceptional appreciation and cyclical exposure make the sustainability of projected earnings more important than the low P/E alone.

SK hynix: HBM Leadership With an Additional Valuation Risk

SK hynix has benefited from its leading position in HBM, particularly its relationships with major AI processor suppliers. Its advanced stacked DRAM technology has allowed the company to capture substantial value from the AI infrastructure expansion, while conventional DRAM and NAND pricing have provided additional earnings support.

The approximately 150% one-year appreciation of its Korean-listed shares reflects these advantages. Nevertheless, the stock’s 31.0% correction suggests that investors are increasingly concerned about competition, capital spending, and the eventual normalization of memory pricing.

Samsung and Micron are pursuing additional HBM capacity, while Chinese manufacturers are becoming more competitive in conventional memory. SK hynix must continue investing in advanced DRAM manufacturing and packaging while also maintaining its NAND business, including manufacturing operations in China.

There is an additional consideration for U.S. investors. Moomoo notes that SKHY can trade at a premium to the underlying Korean-listed shares. That premium introduces a separate valuation risk: even if SK hynix’s operating performance remains strong, changes in the market premium could affect the U.S. security’s return.

SK hynix remains a compelling way to participate in HBM growth, but investors should evaluate both its earnings outlook and the price they are paying relative to its Korean-listed shares. Its technological leadership does not eliminate the risks associated with memory cycles or security-specific valuation premiums.

Segment 2: HDD Storage — Strong AI Demand, but Corrections Approaching 50%

Hard disk drive manufacturers have experienced a substantial recovery as hyperscalers expand storage capacity to accommodate AI datasets, cloud applications, and rapidly growing digital information. Unlike HBM and DRAM, which support high-speed computing, HDDs provide economical mass-capacity storage for data that does not require the same access speeds.

The HDD industry is highly concentrated, with Seagate and Western Digital supplying much of the global market. Increasing storage density, disciplined manufacturing capacity, and demand for enterprise drives have supported pricing and profitability. However, both companies remain exposed to hyperscaler purchasing decisions and the timing of data center expansion.

Seagate Technology: Technology Leadership Supports Growth, but Valuation Matters

Seagate’s 178.5% one-year appreciation reflects the strong recovery in enterprise HDD demand and growing adoption of its high-capacity products. The company’s heat-assisted magnetic recording technology, commercialized through its Mozaic platform, enables greater storage density and improved economics for hyperscale data centers.

Higher-capacity drives allow customers to store more data while limiting increases in power consumption, physical space, and supporting infrastructure. These advantages are particularly valuable as AI applications generate larger datasets and increase long-term storage requirements.

However, Seagate has declined 42.4% from its high, indicating that investors have become more cautious about the sustainability of storage demand and the valuation assigned to future growth. The company’s revenue depends heavily on purchasing decisions by a relatively concentrated group of cloud customers, making it sensitive to changes in deployment schedules and inventory requirements.

Seagate’s technology leadership provides an important competitive advantage, but its substantial stock appreciation means that continued earnings growth may already be reflected in its valuation. The stock becomes more attractive if high-capacity drive demand remains strong and the correction proves greater than any deterioration in earnings expectations.

Western Digital: Similar Growth, but a Much Larger Correction

Western Digital’s 173.4% one-year appreciation closely matches Seagate’s performance, demonstrating that both companies benefited from improving HDD industry conditions. Following the separation of its flash memory business into Sandisk, Western Digital is more directly exposed to HDD demand and the economics of mass-capacity storage.

The company’s high-capacity products serve cloud and enterprise customers seeking economical storage solutions. Increasing AI-generated data and cloud infrastructure expansion support demand, while higher-capacity drives can improve manufacturing economics and customer value.

Western Digital’s 54.0% correction is the largest among the ten companies in Table 1, approximately 12 percentage points greater than Seagate’s decline. This divergence is particularly important because the two companies operate in the same concentrated industry and experienced nearly identical one-year appreciation.

The larger correction raises questions about differences in valuation, product execution, profitability expectations, and investor confidence. It does not necessarily indicate weaker underlying HDD demand, but it suggests that the market is assigning Western Digital greater risk.

For investors seeking exposure to AI-related storage growth, Western Digital’s larger correction may offer greater recovery potential than Seagate’s. However, that conclusion depends on whether its earnings outlook remains comparable to its competitor’s. A larger stock decline is attractive only if the underlying business has not deteriorated proportionately.

Segment 3: Semiconductor Equipment — Are Investors Discounting a Slowdown That Has Not Arrived?

Semiconductor equipment manufacturers occupy a different position in the AI supply chain because their revenues depend on capital expenditure decisions made by semiconductor producers. Applied Materials and Lam Research supply the manufacturing systems required to produce advanced logic, DRAM, NAND, and HBM devices.

Increasing semiconductor complexity creates opportunities for additional equipment spending, but equipment orders can be delayed when customers adjust their capacity expansion schedules. This makes the industry particularly sensitive to expectations about future investment rather than simply current semiconductor demand.

Applied Materials: Broad Technology Exposure, but China and Capital Spending Remain Risks

Applied Materials’ 65.1% one-year appreciation reflects its broad participation in semiconductor manufacturing technology. The company supplies deposition, ion implantation, chemical mechanical polishing, and other systems required for increasingly sophisticated chips.

Advanced transistor structures, HBM, and complex memory architectures require additional manufacturing steps, increasing equipment spending opportunities. Applied Materials benefits from these technology transitions because its equipment participates in numerous stages of semiconductor production.

Nevertheless, the stock has declined 47.8% from its high. Investors are concerned about the timing of future equipment orders, the sustainability of semiconductor capital expenditures, and the company’s exposure to China.

Chinese manufacturers including Naura, AMEC, and Piotech are expanding their capabilities in selected equipment categories, while U.S. export restrictions limit some opportunities for foreign suppliers. These developments create risks that are separate from the global AI investment cycle.

The investment question is whether Applied Materials’ correction has already discounted a substantial reduction in future equipment spending. Its broad technology portfolio supports long-term demand, but near-term performance will depend on customer investment schedules and the company’s ability to maintain competitive positions in major equipment categories.

Lam Research: Record Earnings Challenge the Market’s Pessimism

Lam Research presents one of the most interesting contradictions in the comparison. Its stock gained 56.0% over the past year but subsequently declined 46.9% from its high, including approximately 30% during October according to Moomoo.

Yet the company’s latest operating results demonstrate substantial growth. For the September 2026 quarter, Lam reported revenue of $8.22 billion, up approximately 54% year over year, with non-GAAP gross margin of 52.4% and operating margin of 40.2%.

More importantly, management expects worldwide wafer fabrication equipment spending to increase from the mid-$150 billion range in 2026 to more than $200 billion in 2027, with growth led by DRAM, advanced foundry logic, and NAND. That preliminary forecast implies industry spending growth of roughly 30%, even as Lam’s stock has experienced a severe correction.

This discrepancy suggests that investors are concerned about the sustainability of future growth rather than reacting to an existing collapse in equipment demand. Lam’s approximately 24 times expected fiscal 2027 earnings multiple, as cited by Moomoo, also indicates that the stock is not necessarily inexpensive despite its substantial decline.

Lam remains exposed to China, memory investment cycles, and the timing of new manufacturing capacity. However, its record revenue and strong industry spending outlook challenge the interpretation that the semiconductor equipment market is already entering a downturn.

For investors, Lam offers an attractive long-term position in advanced semiconductor manufacturing, but its valuation and sensitivity to customer capital spending require consideration. The stock’s recovery potential depends on whether management’s optimistic equipment spending outlook materializes.

Segment 4: AI Computing Platforms and Chip Suppliers — Smaller Corrections, but Different Valuations

Nvidia, TSMC, Amazon, and Meta have experienced the smallest corrections in Table 1, ranging from 11.8% to 17.4%. Their businesses differ substantially, but all occupy positions close to AI computing demand, advanced processor manufacturing, or the monetization of AI services.

The relatively modest corrections suggest that investors continue to value their competitive positions and long-term earnings opportunities. However, their one-year returns and valuation characteristics differ considerably, making individual stock selection particularly important.

Nvidia: Relative Stability Reflects Established AI Leadership

Nvidia’s 15.8% one-year appreciation is modest compared with Micron’s extraordinary gain, but its 11.8% correction is the smallest among the ten companies. This combination reflects the timing of investor expectations. Nvidia’s AI leadership was recognized earlier, and its valuation had already incorporated substantial anticipated earnings growth.

The company benefits from its integrated GPU, networking, and software ecosystem, including CUDA, which creates switching costs and reinforces its competitive position. Nvidia remains central to hyperscaler AI infrastructure spending, but its future growth depends on customers continuing to expand computing capacity.

Moomoo estimates that Nvidia trades at approximately 21 times forward earnings, a valuation that appears more reasonable than might be expected for a company with its strategic position. However, the sustainability of AI infrastructure investment remains critical.

Nvidia appears attractive relative to its competitive advantages, but its relatively small correction means investors are receiving less of a valuation discount than in several other AI-related stocks.

TSMC: Manufacturing Leadership Supports Greater Resilience

TSMC’s 44.0% one-year appreciation reflects its dominant position in advanced semiconductor manufacturing. The company produces leading-edge processors for Nvidia, Apple, AMD, and other major chip designers, while its advanced packaging technologies support the integration of AI processors with HBM.

TSMC’s manufacturing capabilities, process yields, and customer relationships create substantial competitive barriers. Unlike equipment manufacturers, whose revenues depend on new capital spending decisions, TSMC generates revenue from ongoing wafer production.

Its 13.9% correction suggests that investors continue to assign significant value to its manufacturing leadership. However, the company remains capital intensive, and its profitability depends on maintaining high utilization rates across expensive fabrication facilities.

TSMC offers exposure to AI processor growth without depending on the success of a single chip designer. Its competitive position supports long-term investment appeal, although its relatively modest correction provides less immediate valuation relief than the larger declines in memory and equipment stocks.

Amazon: Cloud Revenue Provides a Different AI Investment Model

Amazon’s 12.2% one-year appreciation and 12.7% correction place it close to Nvidia and TSMC in the stock performance comparison. However, Amazon’s economic model differs because it purchases and operates AI infrastructure rather than manufacturing semiconductors.

Through AWS, Amazon provides computing services to enterprise customers while developing proprietary processors such as Trainium and Inferentia. Its AI investments create substantial capital requirements, but they also support recurring cloud revenue.

The company’s relative resilience suggests that investors continue to value AWS’s long-term growth potential and Amazon’s ability to distribute infrastructure investment across a broad customer base.

Nevertheless, rising capital expenditures create risks involving utilization, pricing, and returns on invested capital. If AI capacity expands faster than customer demand, Amazon could face pressure on profitability and cash flow.

Amazon remains an attractive long-term AI infrastructure participant, but its relatively small correction suggests that investors are not yet assigning a substantial discount to its growth expectations.

Meta Platforms: The Only One-Year Decliner Offers a Different Recovery Opportunity

Meta is the only company in Table 1 with a negative one-year return, declining 12.45%. Yet its 17.36% correction from its high remains modest compared with the declines experienced by memory, HDD, and equipment manufacturers.

Meta’s AI strategy differs from Amazon’s because it primarily monetizes AI through advertising effectiveness, content recommendations, user engagement, and operating efficiency rather than selling cloud computing services.

The company is investing heavily in AI infrastructure, but the relationship between capital spending and incremental revenue is less direct than for AWS. Investors are increasingly concerned about whether AI investments will generate sufficient returns to justify their cost.

Moomoo estimates that Meta trades at approximately 19 times expected 2027 earnings. That valuation may offer an opportunity if AI improves advertising profitability and capital spending growth eventually moderates.

Meta’s negative one-year return suggests greater investor skepticism than for Amazon, Nvidia, or TSMC. However, its established advertising business provides substantial cash generation, and successful AI monetization could support a recovery.

Meta may offer greater rebound potential than some of its more resilient peers, but that opportunity comes with uncertainty about the timing and profitability of its AI investments.

Investor Takeaway: Which Corrections Offer the Best Opportunities?

The ten-company comparison reveals that AI infrastructure investment is producing distinct stock market outcomes across four business segments. Semiconductor memory manufacturers have generated extraordinary appreciation through shortages and rising prices, but investors remain concerned about the sustainability of those conditions. HDD manufacturers have benefited from increasing data storage requirements, yet their stocks have experienced even greater corrections. Semiconductor equipment companies have delivered strong one-year gains but now face questions about future manufacturing capital expenditures, while AI computing platforms and critical chip suppliers have remained comparatively resilient.

Among the four segments, semiconductor equipment presents one of the more interesting potential recovery opportunities because recent stock corrections contrast sharply with the industry’s projected spending growth. Lam Research’s preliminary forecast for wafer fabrication equipment spending above $200 billion in 2027 suggests that demand could remain strong even as investors reduce equipment company valuations. However, the stocks are not necessarily inexpensive on forward earnings, and any delays in customer spending could produce additional volatility. Memory also offers significant upside potential, particularly Micron, but its reported low earnings multiple must be evaluated against the possibility of a cyclical peak in profitability.

For investors seeking greater stability, Nvidia and TSMC offer strong competitive positions and relatively modest corrections, although their valuations may already reflect substantial future growth. Amazon provides exposure to recurring cloud revenue, while Meta offers potential recovery if AI investments improve advertising profitability and capital spending concerns ease. Seagate and Western Digital remain important beneficiaries of expanding data storage requirements, but their larger corrections demonstrate the risks associated with hyperscaler purchasing cycles and valuation expectations. The most attractive opportunities will ultimately be the companies whose earnings continue growing faster than the market currently anticipates, rather than simply those whose stock prices have declined the most.

Contact [email protected] for any questions or corrections.