share_log

股市动荡之际 “最强阿尔法”浮出水面! 关税重锤压不住井喷式扩张的AI算力

Amidst stock market turbulence, the "strongest alpha" emerges! Tariffs cannot suppress the explosive expansion of AI computing power.

Zhitong Finance ·  Apr 18 16:44

The Lithography Equipment giant$ASML Holding (ASML.US)$Recently, the Earnings Conference showed that management expects both 2025 and 2026 to be years of performance growth, and anticipates that dynamic demand in the end market will shift the chip product mix towards high-end process AI Chips and Datacenter Memory Chips. This semiconductor equipment giant, hailed as the 'pinnacle of human technology,' emphasized that amid tariff pressures, demand for AI computing power remains strong, with the focus of AI computing power shifting towards the seemingly infinite demand for AI inference. This has been confirmed by high-end lithography equipment clients, who have expressed the need for continuous large-scale investments in the field of AI technology.

Holding the title of 'King of Chip Foundry' $Taiwan Semiconductor (TSM.US)$ The earnings data released on Thursday also showed a surge in demand for AI computing power, with Net income skyrocketing by 60%. More importantly, under the heavy pressure of Trump’s tariffs directed at the globe, Taiwan Semiconductor reaffirmed its strong growth data. Taiwan Semiconductor maintains its revenue growth forecast for 2025, expecting this year’s growth rate to still reach around 25%, completely in line with the target set in January, with revenue related to Artificial Intelligence expected to double.

On the tariff-related issues that are drawing global investor attention, Taiwan Semiconductor indicated that it has not observed any changes in customer behavior due to U.S. tariffs, contrasting its optimistic expectations with the uncertainty in the global market. Notably, Taiwan Semiconductor plans to double its advanced CoWoS packaging capacity, mainly for.$NVIDIA (NVDA.US)$The AI GPU production capacity indicates the company's strong confidence that the demand for AI Chips will remain high until early 2026.

During the earnings conference call, TSMC management still expects a compounded annual growth rate target of about 20% for revenues over the next five years (2024-2029), with AI-related revenues expected to grow by around 45%, completely in line with the strong forecasts given at the last earnings meeting, suggesting TSMC has not observed any degree of demand cooling due to Trump's aggressive global tariff policies. Concerning the expectations for the second quarter, TSMC management estimates Q2 revenues to range from $28.4 billion to $29.2 billion, indicating a projected quarter-on-quarter growth of over 13% and a year-on-year growth of over 36%, far exceeding the market expectation of $27.16 billion; it maintains its capital expenditure for 2025 at $38 billion-$-42 billion.

In late March, the largest computer Memory Chip manufacturer in the USA.$Micron Technology (MU.US)$The released Earnings Reports show that, benefiting from the AI infrastructure craze where large enterprises and various government departments are investing heavily in AI, the demand for Memory Chips closely related to AI training/inference systems remains extremely strong, driving a surge in Micron's Datacenter business revenue, including HBM storage systems and enterprise-level SSDs.

Micron's management stated during the Earnings Conference that there is an incredibly strong demand for components of AI infrastructure in data centers for developing iterations and running high-efficiency applications, including so-called 'AI agents'. Coupled with the latest earnings reports and summaries from Taiwan Semiconductor and ASML, this has significantly reinforced the extremely optimistic outlook for the ongoing explosive growth in AI computing power demand. Currently, even the Trump administration's aggressive global tariff policy has not suppressed the continuously surging demand for AI computing power centered around AI chips.

With DeepSeek completely igniting the 'efficiency revolution' in AI training and inference, the future development of AI large models is expected to center fully on 'low cost' and 'high performance'. Under the super wave of AI large models integrating into various industries globally, the demand for cloud AI inference computing power is anticipated to see exponential growth. AI application software (especially generative AI software and AI agents) is rapidly penetrating various industries worldwide, fundamentally transforming the efficiency of business scenarios and significantly increasing sales. The demand for AI computing infrastructure centered around AI GPUs and AI ASICs may present exponential growth, rather than the previously anticipated 'DeepSeek shockwave' triggering a sharp decline in AI computing power demand.

As NVIDIA CEO Jensen Huang mentioned during the February NVIDIA Earnings Conference, the demand for AI chips continues to grow strongly. 'DeepSeek-R1 has ignited global enthusiasm, and the company is excited about the potential demand brought by AI inference. This is an excellent innovation, but more importantly, it has opened up a world-class inference AI model. Models like OpenAI, Grok-3, and DeepSeek-R1 are all inference models that can scale during inference time. Inference models can utilize more than 100 times the computing power.'

According to the latest forecast from the World Semiconductor Trade Statistics (WSTS), the global semiconductor market size is expected to continue growing in 2025 based on 2024's foundation, indicating that the global semiconductor market is likely to grow by approximately 11.2% on top of the already strong recovery trend in 2024, reaching around 697 billion dollars.

WSTS expects that the growth in the semiconductor market size in 2025 will be largely driven by strong demand for enterprise-level memory chips and AI logic chip categories resulting from AI training/inference computing power, with an overall market size for logic chip categories including CPUs, GPUs, and ASIC chips expected to grow by about 17% year-on-year. The market size of memory chip categories covering HBM and enterprise SSDs is expected to grow by more than 13% year-on-year on top of the substantial 81% growth in 2023; at the same time, WSTS also expects that the growth rates of all other segmented chip markets, including Discrete Components, optoelectronics, Sensors, MCUs, and Analog Chips, will reach single-digit increments.

Amidst extreme fluctuations in the global stock market, the 'Alpha' nature of the AI computing power sector stands out.

A recent research report released by Wall Street financial giant Morgan Stanley shows that spending on these two cutting-edge technologies, AI/ML (Artificial Intelligence/Machine Learning), occupies the highest priority in the IT budgets of technology companies in the USA, with demand for security defense software, driven by the AI wave, closely following.

The latest survey and research results published by Morgan Stanley, comprising over ten thousand words, reveal a significant divergence trend in the IT budget expectations of CIOs in the USA, primarily influenced by macroeconomic fluctuations. Although expectations for short-term IT budget growth have declined, confidence in core long-term growth drivers (such as AI and machine learning) and medium-term IT spending remains stable, with expectations for AI spending expected to expand significantly.

As investors further digest and price Trump's radical tariff policies towards the world, the panic sentiment in global stock markets has eased, particularly as technology stocks, which suffered the most severe sell-off in this round of global stock market declines, have shown a brief rebound in prices. Technology stocks are undoubtedly the core driving force for the global stock market entering a long-term bull market in recent years. Therefore, after undergoing a new round of selling, investors are beginning to anticipate that the technology stocks leading the market will sound the counter-offensive horn and lead the global stock market to rebound strongly.

A tweet released last Wednesday during Eastern Time, suggesting a rational shift in the Trump's administration's tariff stance, drove the global stock market, including the US stock market, from ICU to a KTV celebration. Trump stated that he had authorized a 90-day "reciprocal tariff delay" measure for most countries, significantly reducing tariffs on these countries to 10% during this period.

The "tariff tsunami" almost single-handedly created by Trump has resulted in the evaporation of trillions of dollars in global stock market value, and last week it pushed Wall Street traders and global financial market investors to the brink of despair, advising them to "sell everything you can sell." However, after Trump released news on his personal social platform Truth Social about pausing the global "reciprocal tariffs" last week, Wednesday marked an extraordinary rebound in US stocks, leading to an epic surge in global stock markets, while prior to that, over 10 trillion dollars in market value had evaporated.

For the "AI compute industry chain," which has consistently ranked at the top of global stock market investment enthusiasm since 2023, Morgan Stanley's research report shows that the bullish investment logic for AI follows a trend of repair as financial markets begin to refocus on the strong demand expectations for AI chips and other AI infrastructure. After all, compared to the non-AI sectors in semiconductors, the demand expectations for AI GPUs, AI ASICs, HBM, Ethernet Switch chips, and core electrical utilities among the "leaders in AI compute" are much stronger, and they are expected to exhibit the most robust leading trend during tactical rebounds in the market.

Morgan Stanley's latest research indicates that technology stocks closely associated with AI/ML, especially long-standing leaders in the AI computing industry chain—such as NVIDIA, $Broadcom (AVGO.US)$, Micron, and semiconductor giants like Taiwan Semiconductor, are expected to demonstrate significantly stronger "Alpha excess returns" amid the tactical rebound of the US stock market.$S&P 500 Index (.SPX.US)$And has the title of "indicator of Technology Stocks".$NASDAQ 100 Index (.NDX.US)$The so-called "Alpha" is defined as the actual investment returns far exceeding the "Beta returns"—which refers to returns that are significantly higher than those achieved by tracking benchmark stock index investments. The synchronous returns achieved by tracking the benchmark index are also called "Beta returns".$BETA (0263.MY)$)。

Statistics show that in April, the US stock market experienced severe turbulence, but the alpha attributes of AI Chip giants have fully emerged, and the AI computing power industry chain has started to show a leading upward trend. For example, Broadcom, one of the leaders in AI Chips, rose over 2% in April, while the S&P 500 Index fell by as much as 6%, and NVIDIA's GB200 high-speed Copper cable suppliers.$Amphenol (APH.US)$also significantly outperformed the S&P 500 Index. In Asia, SK Hynix, Tokyo Electron Ltd. Unsponsored ADR, and$SMIC (00981.HK)$other AI computing power industry chain leaders all significantly outperformed the Large Cap, and in Europe, ASM International from the Netherlands and BE Semiconductor stocks also greatly outperformed the European stock market.

The investment research platform Seeking Alpha, based on quantitative indicators, has compiled a list of the top ten semiconductor stocks in the USA, indicating that AI chip giants are likely to gain favor from more institutions and retail investors under Trump's "tariff storm." Data compiled by Seeking Alpha shows that these stocks have a market value of at least 10 billion dollars and are ranked according to their proprietary quantitative system. With investors recently increasing their attention on the semiconductor industry, this list provides a data-driven perspective that helps investors understand which companies may occupy the best positions amid the increasingly severe macroeconomic and geopolitical challenges.

For reference, the Seeking Alpha Algo system is driven by SA's unique "Quant System" which rates stocks based on their overall value, growth potential, profitability, EPS revision trends, and price momentum indicators. The leaderboard shows that the top two positions are held by the two leading AI Chip companies, with AI ASIC chip leader Broadcom (AVGO.US) ranking first, and AI GPU leader NVIDIA (NVDA.US) in second place. The following image shows the top ten US semiconductor stocks.

Morgan Stanley and other major Wall Street firms continue to focus on repairing AI investment logic.

On the eve of the earnings season, Wall Street's well-known investment firm Oppenheimer reiterated its preference for the "three giants of AI Chips," namely NVIDIA (NVDA.US), Broadcom (AVGO.US), and $Marvell Technology (MRVL.US)$recommended them as top picks in the Semiconductor Sector. The Oppenheimer analyst team wrote in a report to clients: "In the chaotic macro environment and tariff context, we believe AI Chips are the strongest and safest growth direction."

A recent research report from KeyBanc Capital Markets shows that the global semiconductor industry is presenting a "divided pattern"—with strong demand for AI chips while other types of chips are still struggling to recover from a downturn. KeyBanc's analysts stated that the strongest demand theme for AI chips is still fully dominated by NVIDIA, which occupies 80%-90% of the AI chip market share, while other AI chip participants generally have not escaped negative catalysts. The analysts specifically pointed out that the mass production progress of the Blackwell architecture AI GPU is proceeding smoothly, and the demand for advanced CoWoS packaging remains stable. "The demand for AI chips continues to grow explosively, and we also see multiple cross-level influences in the field of NVIDIA AI GPU and ASIC, which are AI-specific chips," KeyBanc stated.

UBS, an international major bank, released a research report stating that although it is difficult to assert that semiconductor giants can completely avoid the destruction of semiconductor demand related to Trump's aggressive tariff policy, UBS firmly believes that AI spending will remain resilient. The overall weak demand environment may prompt companies to accelerate the adoption of generative AI technologies to reduce operating costs, therefore UBS will focus more on AI computing power-driven stocks, such as NVIDIA and Broadcom.

Morgan Stanley's research report shows that, $Dell Technologies (DELL.US)$Among enterprise Hardware suppliers, performance is outstanding, with net spending intentions reaching +35%, up 1 percentage point from Q3 2024, setting a new high level in Morgan Stanley's survey reports over the past eight years, reflecting that major technology companies in the USA still have very strong demand for the most core hardware in the AI infrastructure field. Morgan Stanley's survey data indicates that as high as 55% of CIOs who are assessing or planning to assess AI technology show a tendency to allocate towards Dell's AI infrastructure solutions, slightly up from 54% in Q3 2024. This survey result also indicates that the demand for high-performance AI servers equipped with AI GPUs and AI ASICs remains very strong.

In terms of enterprise AI computing power expenditure, USA tech giants have also started to join Wall Street—namely, the extremely bullish line regarding the expansion of AI computing resource demand, such as Alphabet-A. $Alphabet-A (GOOGL.US)$CEO Sundar Pichai reiterated the tech giant's recently committed investment of approximately 75 billion USD for the construction of a large-scale AI infrastructure investment plan for AI data centers.

$Amazon (AMZN.US)$Leader Andy Jassy unusually defended the company's massive investment in the field of AI in a high-profile manner in the annual public letter to shareholders last week. The helm of this USA e-commerce giant candidly stated, "When every customer experience will be reshaped by AI, deeply laying out AI is not a choice but a survival question." Amazon's management still expects capital expenditure to reach 100 billion USD this year, most of which will be allocated to AI-related infrastructure projects.

Deutsche Bank, UBS Group, and Piper Sandler have slightly lowered their target price for NVIDIA in the latest Research Reports, but still maintain at least a 12-month target price of $135, along with the most optimistic rating equivalent to "Buy." The latest ratings and target prices compiled by TIPRANKS show that Wall Street's consensus rating for NVIDIA stocks is "Strong Buy," with no "Sell" ratings appearing; the average target price within 12 months is as high as $170, indicating a potential upside of up to 68%; the highest target price reaches $200, while the most pessimistic target price is still significantly above the current stock price at $120.

Editor/danial

The translation is provided by third-party software.


The above content is for informational or educational purposes only and does not constitute any investment advice related to Futu. Although we strive to ensure the truthfulness, accuracy, and originality of all such content, we cannot guarantee it.
    Write a comment