Memory Shock, Middle East Fear, Rumor Selloff

● Memory Shock, Middle East Fear, Rumor Selloff

Why Micron Rose While Samsung Electronics and SK Hynix Fell: Key Drivers Were Middle East Risk, Memory Price Resistance, and the Spread of False Signals

This decline cannot be explained simply as a broad weakness in semiconductors.

In the U.S. market, Micron and SanDisk advanced, while Samsung Electronics and SK Hynix weakened in Korea.

At first glance, this appears inconsistent, but the market was reacting to three specific factors.

First, rising tensions in the Middle East pushed crude oil sharply higher, weakening sentiment in both the Nasdaq and the KOSPI.

Second, Chinese smartphone makers Oppo and Vivo reportedly signaled resistance to elevated memory prices, effectively indicating a buying boycott.

Third, an old negative memory-sector report associated with Morgan Stanley spread through Telegram as if it were a current development, intensifying forced selling.

In other words, this move is better understood as the result of price resistance, geopolitical risk, and rumor-driven trading rather than a slowdown in AI semiconductor demand.

1. U.S. Market Backdrop: Big Tech Was Weak, but Memory Stocks Held Up

The U.S. market provides the initial context.

Escalating tensions in the Middle East raised concerns that crude oil could move above $100 per barrel.

A surge in oil prices quickly translates into renewed inflation concerns.

Higher inflation expectations reduce the likelihood of near-term Fed rate cuts and pressure growth stocks, especially in the Nasdaq.

As a result, large-cap technology stocks broadly weakened in the prior U.S. session.

Middle East risk increases – oil prices rise
Higher oil prices – renewed inflation concerns
Inflation concerns – weaker expectations for rate cuts
Higher rate pressure – downside pressure on Nasdaq growth stocks

However, there was an important exception.

Micron and SanDisk rose instead.

The reason was the market’s continued expectation that Big Tech would keep purchasing memory products.

AI servers, data centers, HBM, and high-performance SSD demand remained sufficient to support the view that memory earnings could continue to improve.

As a result, sentiment toward Samsung Electronics and SK Hynix was not yet fully broken at the start of the Korean session.

2. Why Korean Semiconductor Shares Fell: The First Negative Catalyst Was a Chinese Smartphone Price Boycott Signal

The first factor weighing on domestic semiconductor sentiment was activity from Chinese smartphone makers.

Reports circulated that Oppo and Vivo were effectively delaying purchases or signaling a boycott because memory prices had become too high.

This point is important.

Until now, the memory market had been supported by supply shortages and expectations of AI-driven demand.

But when end customers signal that they are unwilling to buy at current prices, the market begins to question the sustainability of the move.

Have memory prices risen too quickly?
Are smartphone makers facing renewed inventory pressure?
Is AI server demand strong while mobile demand remains weak?
Could the DRAM and NAND price upcycle be shorter than expected?

This does not necessarily imply a collapse in the memory cycle.

Given that supply constraints remain in place, this may also reflect buyer pressure in price negotiations.

However, equity markets react to speed before they react to facts.

For large-cap Korean names such as Samsung Electronics and SK Hynix, earnings sensitivity to memory pricing is high.

Accordingly, reports that Chinese smartphone makers were refusing to buy at higher prices became an immediate selling catalyst.

3. Second Negative Catalyst: Morgan Stanley Memory Bearish Commentary Was Circulated as If It Were a New Warning

The second factor intensifying the decline was a rumor linked to Morgan Stanley.

Messages spread rapidly through Telegram and other channels suggesting that analyst Shawn Kim at Morgan Stanley was negative on the memory outlook.

The key issue was that this material was not a new report, but an older note from 2022.

In other words, outdated content was redistributed as if it reflected current conditions.

This matters because the market was already in a highly sensitive state.

Investors were simultaneously watching Middle East risk, rising oil prices, Nasdaq weakness, exchange-rate volatility, and concerns about a semiconductor peak.

In that environment, any message implying that a major foreign brokerage was bearish on memory quickly triggered defensive selling.

Selling began before verification.
Algorithmic traders and short-term participants followed the downside move.
Retail investors considered cutting losses.
Weaker foreign flows amplified the decline.

This rumor amplification effect likely played a meaningful role in the decline of Samsung Electronics and SK Hynix.

Given the heavy weighting of semiconductors in the Korean market, weakness in these large-cap names also pressured the broader index.

4. Why Micron Rose While Samsung Electronics and SK Hynix Fell

Although these are all memory semiconductor companies, the market was pricing them differently.

Micron is viewed more directly as a beneficiary of U.S. AI infrastructure spending.
Samsung Electronics and SK Hynix are exposed not only to AI demand but also to Chinese mobile demand, exchange rates, and Korean market flows.
The U.S. market focused on whether Big Tech would continue buying memory.
The Korean market was more sensitive to Chinese smartphone pricing resistance and foreign-media rumor risk.

SK Hynix, in particular, has been one of the primary HBM beneficiaries, so valuations had already moved significantly higher and became more sensitive to adverse news.

Samsung Electronics has been priced on both memory recovery and expectations for foundry improvement, but in the short term it remains highly sensitive to DRAM pricing and mobile demand signals.

For that reason, Micron’s rise in the U.S. does not necessarily imply that Samsung Electronics and SK Hynix must move in the same direction in Korea.

5. Market Interpretation: The Memory Supercycle Has Not Ended; the Market Has Entered a Pricing Negotiation Phase

The most important interpretation is this.

The reported boycott-like behavior by Chinese firms does not mean memory demand has disappeared.

It is more likely a sign that buyers are pushing back as prices rise too quickly.

Memory is a cyclical industry.

When prices are too low, suppliers cut output; when prices rise, customers delay purchases.

At present, AI server demand for high-end memory remains strong, while smartphones and PCs are starting to show resistance to higher prices.

HBM demand for AI servers remains strong.
Demand for data-center SSDs and high-performance DRAM remains firm.
Mobile memory customers are beginning to resist higher pricing.
Going forward, the key issue is no longer whether memory is broadly strong, but which type of memory is strongest.

Investors should therefore assess Samsung Electronics, SK Hynix, and Micron by product mix rather than treating them as a single group.

HBM, DDR5, LPDDR, NAND, and eSSD may no longer move in lockstep.

6. The Key Point Often Missed Elsewhere: This Selloff Was Driven More by Information Asymmetry Than by Demand Deterioration

The most important issue here is not memory pricing itself, but how information spread through the market.

Older research was redistributed as if it were a new negative signal, while a price-sensitive move by Chinese buyers was interpreted as evidence of a broader downturn.

This is a classic information-asymmetry environment.

Today’s market often reacts before investors verify the underlying news.

Especially when bearish messages spread quickly through Telegram, forums, and short-term trading channels, prices can move before the facts are confirmed.

In such a setting, near-term performance is often driven more by who sold first than by fundamental changes.

Middle East risk is a genuine macro headwind.
Chinese smartphone pricing resistance requires close monitoring.
However, recycling a 2022 note as if it were current is a market distortion.
These three factors together intensified semiconductor share weakness.

Accordingly, it is premature to conclude that the AI semiconductor cycle has ended.

Rather, this episode shows how quickly the market can punish perceived risk in an overheated segment.

7. What Investors Should Monitor Now

When evaluating Samsung Electronics and SK Hynix, investors should track the following indicators alongside share prices.

HBM supply contracts: linked to Nvidia, AMD, and hyperscale data-center investment.
DRAM spot and contract pricing: a key indicator of the strength of the memory recovery.
NAND price stability: also relevant for Micron and SanDisk.
China smartphone shipment trends: critical for mobile memory demand from Oppo, Vivo, and Xiaomi.
Crude oil and exchange rates: directly affect foreign flows into Korean equities and overall valuation support.

If oil prices remain elevated, the pressure will extend beyond semiconductors to the broader equity market.

Higher oil prices increase inflation pressure and may delay expectations for U.S. rate cuts.

A slower rate-cut cycle would weigh on valuations for growth and technology stocks.

In short, semiconductors must now be viewed through both earnings and macro variables.

8. Company-Specific View: Samsung Electronics and SK Hynix

Samsung Electronics is currently supported by expectations for memory recovery, HBM competitiveness, and improvement in foundry operations.

However, in the near term it remains highly sensitive to mobile DRAM and NAND pricing.

Reports of pricing resistance from Chinese smartphone makers can affect Samsung Electronics across a broader range of businesses.

SK Hynix is the name with the strongest HBM exposure.

It remains a core supplier in the AI semiconductor market, so the long-term growth story remains intact.

That said, after a strong share-price run, even modest negative news can trigger profit-taking.

Micron is viewed in the U.S. market as a key beneficiary of AI infrastructure spending and memory recovery.

U.S. investors tend to focus more directly on the durability of Big Tech’s AI server investment cycle.

This is why the same memory-sector news produced different market reactions in the U.S. and Korea.

9. Conclusion: This Was a Warning on Market Sensitivity, Not a Signal That the Memory Cycle Has Broken

The weakness in Samsung Electronics and SK Hynix cannot be explained by a single factor.

Middle East tensions weakened global risk sentiment, and rising oil prices pressured Nasdaq sentiment.

Chinese smartphone pricing resistance added a second negative catalyst, while an old Morgan Stanley note was circulated as a current bearish signal, amplifying selling pressure.

However, it is too early to describe this as the end of the memory semiconductor cycle.

AI servers and data-center demand remain structurally strong.

The more likely outcome is a market in which product-level competitiveness and customer-specific demand separate the winners from the rest.

Investors should focus on distinguishing between real deterioration, pricing negotiation, and rumor-driven volatility.

Failing to make that distinction can lead to selling strong companies at the wrong time or misreading a normal correction as a structural change.

< Summary >

Micron rose on expectations tied to U.S. AI infrastructure demand, while Samsung Electronics and SK Hynix fell on reports of Chinese smartphone price resistance and the spread of an outdated Morgan Stanley rumor.

Middle East risk and the surge in oil prices also weakened sentiment in both the KOSPI and the Nasdaq.

This decline is better viewed as a short-term shock caused by pricing negotiation, geopolitical risk, and rumor amplification rather than a collapse in the memory semiconductor cycle.

Going forward, investors should differentiate between HBM, DRAM, NAND, and mobile demand, with the key question remaining whether AI semiconductor demand is actually slowing.

[Related Articles…]

*Source: [ 내일은 투자왕 – 김단테 ]

– 마이크론 올라도 전자닉스 떨어진 이유 #삼성전자 #하이닉스 #마이크론

● CPU, AI, Surge

Is the Next Battleground After Memory CPUs? Top 4 U.S. CPU Companies to Watch in the Age of Agentic AI

The key shift in AI investing is moving from a market focused only on GPUs to one that must also consider CPUs.

In the early phase of generative AI, Nvidia GPUs and HBM memory were the primary beneficiaries. In the agentic AI era, however, the role of CPU semiconductors may become significantly more important as systems interpret instructions, form plans, connect multiple applications and data sources, and coordinate execution.

This report reviews why the next battleground in the AI semiconductor cycle may be CPUs, the signals visible in Stanley Druckenmiller’s 13F filing, the distinctions among Intel, AMD, ARM, and Qualcomm, and how to access the theme through a U.S. semiconductor ETF.

The key point often overlooked in other coverage is that CPU demand may be driven not only by PC replacement cycles, but also by its role in resolving bottlenecks in the execution layer of AI agents.

1. News Summary: The AI Investment Cycle Is Expanding from GPU to CPU

The market trend is straightforward.

From 2023 through 2025, the center of the AI semiconductor market was Nvidia GPUs, HBM memory, and data center expansion.

As AI evolves beyond content generation and into agentic AI that performs real tasks, the importance of CPU semiconductors is rising again.

Generative AI phase: The focus was on training large datasets and generating responses quickly.
Inference AI phase: The focus shifted to more accurate and faster answers to user prompts.
Agentic AI phase: Systems must handle planning, tool calls, data retrieval, payments, reservations, and execution coordination.

GPUs are strong at high-volume parallel computation, but CPUs are responsible for sequencing complex instructions and coordinating multiple tasks.

In that sense, the center of gravity in AI semiconductor investing is expanding from GPUs and memory to CPUs.

2. Why CPUs Matter Now: How Agentic AI Is Changing Semiconductor Demand

Consider a user asking an AI to “plan a trip to Tokyo.”

Traditional generative AI would typically provide travel recommendations, a sample itinerary, and a list of restaurants.

Agentic AI is different.

It compares airfare prices.
It searches hotel inventory across lodging apps.
It calculates transportation routes using mapping services.
It revises the itinerary based on the user’s budget.
If needed, it connects to reservation and payment steps.

As AI begins to invoke multiple services, exchange data, recalculate conditions, and execute outcomes, CPU performance becomes central to workflow orchestration.

If GPUs are the “fast computation engine,” CPUs are the “system coordinator.”

Accordingly, future AI performance analysis will need to examine not only GPU speed, but also how efficiently CPUs allocate tasks and reduce bottlenecks.

3. A Signal from Druckenmiller’s 13F Filing: Partial Big Tech Sales, New CPU Exposure

One notable development is the portfolio shift of Stanley Druckenmiller.

Druckenmiller, known for his work with George Soros, is widely recognized as a global macro investor with a long record of strong returns.

He is also sometimes referred to as a “shadow economic president.”

This reflects not only his investment performance, but also his deep network among key figures in U.S. economic policy.

Recent 13F disclosures highlight the following:

He reduced positions in Alphabet and Amazon.
He initiated a new position in Intel.
He initiated a new position in Arm Holdings.
Broadcom also appeared among the new additions.

While the filing date and current holdings may differ, the direction of the portfolio is still meaningful.

Intel and Arm Holdings are both central to the CPU semiconductor market.

Based on the filing, Intel was cited at roughly KRW 27 billion and Arm Holdings at roughly KRW 24 billion, suggesting that AI infrastructure exposure is broadening beyond GPUs to CPUs and custom semiconductors.

4. CPU vs. GPU: The Core Distinction Investors Need to Know

The technical details do not need to be overly complex.

From an investment perspective, understanding the functional difference between CPUs and GPUs is sufficient.

CPU: Processes complex instructions sequentially and coordinates the overall system.
GPU: Performs large volumes of simpler operations in parallel.
HBM memory: High-performance memory that enables GPUs to read and write data faster.
AI data center: A large-scale AI infrastructure in which GPUs, CPUs, memory, and networking equipment operate together.

During the early phase of generative AI, the rapid expansion in training workloads made GPUs and HBM the key bottlenecks.

As AI moves into task execution, CPUs are expected to play a larger role in connecting tools and data sources.

For that reason, CPU demand should be viewed not merely as PC replacement demand, but as structural demand tied to AI data centers and the spread of agentic AI.

5. Important Evidence: TrendForce’s “1-to-1 Theory” and Nvidia’s Entry into CPUs

Market commentary increasingly suggests that the demand ratio between CPUs and GPUs in AI inference and agentic AI environments may become much more balanced than in the past.

In the training phase, GPU usage was overwhelmingly dominant.

For example, a structure of 8-to-1 or 10-to-1 in favor of GPUs may have been plausible. In inference environments where AI agents execute actual tasks, some analysis suggests the ratio could move closer to 1-to-1.

That is the core reason CPUs warrant renewed attention.

It is also notable that Nvidia, the leading GPU company, is strengthening its CPU strategy.

Nvidia introduced the Grace CPU.
It later expanded the roadmap with Vera CPU.
Jensen Huang has also noted that CPUs may become more important for web database queries, tool calls, and AI execution workflows.

In other words, the leading GPU company itself is signaling that CPUs matter more.

This should be viewed less as a short-term theme and more as a sign of structural change in AI semiconductors.

6. Intel’s Earnings Indicate a Shift: More AI CPU Contracts

Intel has long been viewed as a lagging name in U.S. equities.

Compared with Nvidia, AMD, and TSMC, investors questioned its growth profile, and uncertainty around its foundry business remained a concern.

However, recent earnings highlighted an important development.

AI-related CPU contracts increased.
An LTA agreement was cited as contributing approximately $5.1 billion in revenue.
Intel still retains a strong ecosystem in the server CPU market.

Intel is not simply a CPU company.

It is also a strategically important company in the U.S. effort to rebuild semiconductor supply chains.

Given the U.S. policy direction toward reshoring manufacturing capacity, Intel’s foundry business remains an important variable for both the semiconductor sector and the broader economic outlook.

7. ARM’s Strategic Shift: From IP Licensing to Direct CPU Chips

Arm Holdings is also undergoing an important transition.

Historically, ARM has been known less as a direct chip manufacturer and more as a provider of CPU design IP and a recipient of licensing fees.

Apple, Qualcomm, Nvidia, and many others use ARM architecture.

At recent events, ARM signaled that it intends to move beyond design and expand into direct chip development.

Its plan to build CPUs suited to inference AI is particularly important.

Historically, IP licensing was the core business model.
Going forward, direct entry into the CPU chip market may broaden the opportunity set.
ARM-based CPUs may gain traction in inference and low-power server markets.

ARM functions as a provider of the design language for the semiconductor ecosystem.

As companies using ARM-based chips grow, ARM’s royalty revenue may increase as well.

8. The Two Main CPU Architectures: x86 and ARM

The most important distinction in CPUs is between x86 and ARM architectures.

In simple terms, architecture refers to the way a CPU is designed and the instruction set it uses.

8-1. x86 Architecture: Intel and AMD’s Domain

x86 is the architecture that has long dominated the traditional PC and server markets.

It began with Intel’s 8086 CPU in 1978 and later evolved through the 80186, 80286, 386, 486, and 586 generations. The “86” in the name became the basis for the x86 label.

x86 is strong in handling complex instructions.
It has a highly established PC and server ecosystem.
Intel and AMD are the main companies in the space.
Software compatibility is a major advantage.

Intel led the CPU ecosystem from the IBM PC era, while AMD grew by competing within the same x86 framework.

Today, Intel and AMD remain core players in the server CPU market.

8-2. ARM Architecture: A Leader in Low Power and Custom Design

ARM has a strong advantage in mobile and low-power devices.

The architecture is now extending from smartphones and tablets to laptops and, increasingly, data center servers.

It offers strong power efficiency.
It is well suited to custom design.
It is used in Apple’s M-series, Qualcomm’s Snapdragon, and Nvidia’s Grace CPU.
It is favored by companies building custom chips.

In AI data centers, power cost is a major issue.

When GPUs and CPUs are deployed at scale, efficiency can directly affect profitability.

For that reason, ARM architecture may continue to attract attention in U.S. semiconductor ETFs and AI infrastructure strategies.

9. Top 4 U.S. CPU Companies Investors Should Watch: Intel, AMD, ARM, Qualcomm

From an investment standpoint, the number of core CPU names is limited.

Among U.S.-listed companies, Intel, AMD, Arm Holdings, and Qualcomm are the most important names in the sector.

9-1. Intel: A Core Part of the x86 Ecosystem and U.S. Semiconductor Supply Chains

Intel still has a strong position in the server CPU market.

Its Xeon CPUs have long been used in data centers and enterprise servers, with a clear advantage in software compatibility and ecosystem depth.

Intel is also one of the few companies with domestic foundry capacity in the United States.

As the U.S. seeks to reconfigure semiconductor supply chains around domestic capabilities, Intel’s strategic value extends beyond its financial results.

9-2. AMD: A Strong Competitor Designing Both CPUs and GPUs

AMD belongs to the same x86 camp as Intel, but it has rapidly gained share in the server CPU market.

Its EPYC CPUs are well regarded for core count, power efficiency, and pricing.

AMD’s key advantage is that it designs both CPUs and GPUs.

In AI systems, a bottleneck can emerge when data moves from the CPU to the GPU.

Because AMD can design both components, it can optimize around that bottleneck more effectively.

9-3. Arm Holdings: A Design IP Platform for the AI Inference Era

ARM remains primarily a design IP and licensing business rather than a manufacturer.

However, by signaling direct entry into the inference CPU market, it is expanding its growth narrative.

As companies using ARM-based chips, including Apple, Nvidia, and Qualcomm, continue to grow, ARM’s royalty revenue may also rise.

If demand for low-power, high-efficiency CPUs increases in the AI agent era, ARM may benefit structurally.

9-4. Qualcomm: Expanding from Mobile CPUs into PCs and Servers

Qualcomm established its position in mobile processors through Snapdragon.

In battery-powered devices such as smartphones, power efficiency is critical, and Qualcomm has demonstrated competitiveness in this area for years.

More recently, it has been pursuing expansion into the PC and server markets through its Oryon CPU.

In other words, Qualcomm may evolve from a mobile semiconductor company into a player in AI PCs and low-power server CPUs.

10. ETF Strategy: KIWOOM U.S. CPU Semiconductor TOP 4+

If selecting individual stocks is difficult, investors may consider gaining exposure through a CPU semiconductor ETF.

The product highlighted here is the KIWOOM U.S. CPU Semiconductor TOP 4+ ETF.

The expected listing date was cited as July 28.

The ETF concentrates roughly 80% of assets in the four leading U.S. CPU semiconductor names, while the remaining 20% is allocated to related ecosystem companies.

Main exposure: U.S. CPU semiconductor companies
Number of holdings: 10
TOP 4 weight: Approximately 80%
Remaining weight: Includes related names such as Nvidia, Apple, and TSMC

The reported weights are as follows:

AMD: approximately 25.62%
Intel: approximately 24.15%
Qualcomm: approximately 15%
Arm Holdings: approximately 14%

AMD and Intel alone account for nearly 50% combined.

Including Qualcomm and Arm Holdings reinforces the ETF’s concentrated exposure to major CPU semiconductor names.

As with all ETF strategies, actual holdings and weights may change over time, so investors should review the fund manager’s materials before investing.

11. The Key Advantage of This ETF: Exposure to Both x86 and ARM

The main advantage of this ETF is that it includes both x86 and ARM architecture exposure.

From an investor’s perspective, it is difficult to know in advance whether Intel, AMD, or ARM-based CPUs will gain the strongest adoption.

For that reason, investing in the overall growth of the CPU semiconductor market may be more practical than selecting a single winner.

x86 is represented by Intel and AMD.
ARM is represented by Arm Holdings and Qualcomm.
Nvidia and Apple also use ARM-based chips.
TSMC is closely linked to the semiconductor manufacturing ecosystem.

The AI semiconductor market is not a single-stock story. It is a broad infrastructure cycle involving CPUs, GPUs, memory, foundries, and networking.

Accordingly, a U.S. semiconductor ETF can provide diversified exposure while reducing single-name risk.

12. The Core Point Other Coverage Often Misses

The most important point is not that CPUs are a new theme, but that the bottleneck in AI systems is shifting.

Until now, the main constraint in AI has been GPUs and HBM, which are required for training large models.

That is why companies such as Nvidia, SK Hynix, Micron, and TSMC attracted significant attention.

However, as AI moves into agentic execution, the bottleneck may shift from raw computation to coordination and execution.

Once AI begins searching the web, querying databases, calling multiple apps, and connecting to payment systems, the role of the CPU expands.

In that sense, CPU semiconductors are tied to a structural shift that may be larger than a simple recovery in the PC market.

Missing this point leads to an overly GPU-centric interpretation of the AI investment cycle.

The next phase is not that GPUs are finished, but that the investment framework is expanding to include CPUs as well.

13. Risks Investors Should Monitor

Even if the growth potential of the CPU semiconductor market is strong, the risks are clear.

Concentration risk: High weights in the top four holdings may cause ETF performance to depend heavily on a few names.
Technology competition risk: It remains uncertain whether x86 or ARM will scale faster.
Valuation risk: Many AI semiconductor stocks already reflect elevated expectations.
Currency risk: U.S. stocks and ETFs are affected by KRW/USD exchange rate movements.
Policy risk: Semiconductor supply chain policy, China-related restrictions, and U.S. industrial policy may affect earnings.

Rather than chasing short-term momentum, a phased approach within the broader themes of AI data centers, agentic AI, and U.S. semiconductor supply chains may be more appropriate.

14. Who This May Be Suitable For

Funds such as the KIWOOM U.S. CPU Semiconductor TOP 4+ ETF may be suitable for investors who:

Want long-term exposure to AI semiconductor growth
Are looking for the next investment theme after Nvidia and HBM
Find it difficult to choose among Intel, AMD, ARM, and Qualcomm
Prefer U.S.-listed semiconductor ETF exposure
Already have a high domestic semiconductor allocation and want to diversify

For investors focused on short-term thematic trading, or those unable to tolerate high volatility, position sizing should be managed carefully.

AI semiconductors may offer long-term growth, but price volatility can remain high.

15. Conclusion: The Next Phase After Memory May Be CPUs

The AI investment trend has already gone through one major shift.

It began with generative AI software, then moved to Nvidia GPUs and HBM memory.

Now, as AI advances into agentic execution, CPU semiconductors are emerging as the next battleground.

Intel and AMD remain anchored in the x86 ecosystem and continue to defend the server CPU market.

ARM and Qualcomm are expanding from low-power and custom design into mobile, PC, and server opportunities.

Nvidia is also emphasizing the importance of CPUs, while ARM has signaled possible direct chip-market entry.

Ultimately, AI semiconductor investing may evolve from a GPU-only framework to a broader infrastructure framework that includes CPUs, GPUs, memory, and foundries.

< Summary >

The next battleground in AI investing may be CPU semiconductors.

In the generative AI phase, GPUs and HBM were the key beneficiaries, but in the agentic AI era, CPUs are becoming more important for coordination and execution.

Druckenmiller’s 13F filing also showed new positions in Intel, Arm Holdings, and Broadcom, reinforcing interest in CPU-related names.

The CPU market is divided between the x86 camp, led by Intel and AMD, and the ARM camp, led by ARM and Qualcomm.

The KIWOOM U.S. CPU Semiconductor TOP 4+ ETF concentrates roughly 80% of assets in AMD, Intel, Qualcomm, and Arm Holdings.

The key point is not that GPUs have lost relevance, but that the AI semiconductor opportunity set is expanding to include CPUs.

[Related Articles…]

*Source: [ 소수몽키 ]

– 메모리 다음 차례는 여기? AI 투자 다음 격전지 승자가 될 주식들

NextGenInsight.Net

Like this: