芯片公司财报季拉开帷幕，“极端化” 席卷行业：AI 芯片需求炸裂，非 AI 全线萎靡

According to the Zhitong Finance APP, the new round of financial reports from the two core forces in the global chip industry - ASML (ASML.US) and Taiwan Semiconductor (TSM.US) - shows that in this global high-end industry chain worth as much as $530 billion, the disparity between the stock prices and actual performance of chip companies riding the unprecedented wave of artificial intelligence and those that have failed to catch this AI trend is growing. This highlights the market's preference for betting real money on listed chip companies linked to AI rather than those whose main business is not fully tied to the AI field. Based on preliminary data from this financial reporting season, this gap between stock prices and performance may soon widen from a crack to an abyss.

"Without artificial intelligence, the entire chip market would be very difficult." Christophe Fouquet, CEO of the Dutch lithography machine manufacturer ASML, said during last week's earnings conference call. The company unexpectedly fell short of expectations in its Q3 performance last week, mainly due to the continued softness in all chip demand outside of artificial intelligence, leading to a downward revision of its full-year sales forecast for 2025. ASML's Q3 orders also fell significantly below market expectations due to sluggish demand in non-AI areas.

ASML, based in the Netherlands, is the world's largest lithography system manufacturer, and its lithography equipment plays a crucial role in the chip manufacturing process. ASML is the sole supplier of the most advanced extreme ultraviolet (EUV) lithography machines currently used by TSMC, Samsung, and Intel to manufacture high-end process chips.

If chips are the "pearl" of modern human industry, then lithography machines are the tools necessary to produce this "pearl". More importantly, ASML is the only supplier in the world of EUV lithography equipment needed for the most advanced process chips such as 3nm, 5nm, and 7nm chips. The company's downward revision of performance expectations reflects that the demand level in the entire chip industry is still in a weak recovery phase, at least not achieving a "boom cycle" of chip industry demand at this time.

Chip Industry's "Two Worlds": Surging demand for AI-related chips, while non-AI remains in a "sluggish moment"

ASML's latest performance announcement has raised new concerns about the health of the chip industry. Despite the unprecedented AI boom, the demand for chips closely related to AI, especially data center AI chips, has surged. U.S. tech giants continue to invest heavily in AI GPUs, data center Ethernet chips, and other artificial intelligence infrastructure to expand or build new data centers. However, the entire chip industry is still significantly impacted by weak demand in key markets such as personal computers, smartphones, and electric vehicles.

Furthermore, this lithography giant has long been caught in the escalating geopolitical tensions between the U.S. and China, which may further cut off more channels for ASML lithography machines to enter the Chinese chip market, thereby continuing to impact ASML's performance. The Chinese market remains the world's largest market for semiconductor equipment such as lithography machines From the perspective of rational investors, ASML's performance in hitting the global chip stocks does not mean that the global frenzy of AI chip deployment is retreating or cooling down, and ASML's performance shows a continuous surge in demand for AI chips. However, this explosive financial report does reveal the latest dynamics of the global chip industry, that is: the AI frenzy is still ongoing, especially the demand for all types of AI chips focusing on B-end data centers remains very hot. However, areas unrelated to AI, such as electric vehicles, industrial sectors, IoT devices, and a wide range of consumer electronics chip demands, are still in a state of soft demand or even a significant decline in demand.

However, last Thursday, the chip manufacturing giant Taiwan Semiconductor (TSM.US), known as the "king of global chip foundries," significantly raised its revenue forecast for 2024 and emphasized the crazy demand for AI after the management highlighted the demand for AI chips. This greatly eased the market's concerns about the overall chip industry demand and greatly boosted the market's optimism about the demand for AI chips. Apple and the AI chip superpower NVIDIA are both core chip manufacturing customers of Taiwan Semiconductor. Taiwan Semiconductor's CEO C.C. Wei stated at the earnings conference that industry growth is being strongly driven by AI-related factors, and overall chip demand is "stable" and showing signs of improvement.

Taiwan Semiconductor's latest performance has significantly reinforced the investment view that the AI frenzy is still in full swing, and the demand for AI chips remains extremely hot. When talking about the market demand for AI chips, C.C. Wei, the head of Taiwan Semiconductor, stated at the earnings conference that the outlook for AI chip demand is very optimistic, and emphasized that customers of Taiwan Semiconductor have a far greater demand for CoWoS advanced packaging than the company can supply.

"The company will fully respond to customers' demand for CoWoS advanced packaging capacity. Even if production capacity doubles this year and continues to double next year, it is still far from enough," C.C. Wei said at the earnings conference. CoWoS advanced packaging capacity is crucial for more extensive AI chips such as NVIDIA's Blackwell AI GPU. "Almost all AI innovators collaborate with Taiwan Semiconductor, and the demand for AI-related products is real, and I believe this is just the beginning."

Taiwan Semiconductor's management expects the company's full-year revenue to grow by nearly 30%, exceeding the general analyst expectation of 20%-25% and the company's guidance from the previous quarter. The management also expects the revenue related to data center AI server chips (including NVIDIA AI GPU, Broadcom AI ASIC, and other broadly defined AI chips) at Taiwan Semiconductor to more than double this year.

The successive release of financial reports by ASML and Taiwan Semiconductor marks the beginning of a new round of global chip company earnings season, and the performance of these two core giants in the chip industry chain jointly shows that the global AI spending wave is unstoppable. The logic supporting stocks closely related to AI chips can be described as extremely robust, and the rising trend of leaders in AI chips such as NVIDIA may be far from over. Especially for NVIDIA, the AI chip leader with a market share of 80%-90% in the data center AI chip field, the stock price may continue to hit historical highs, breaking through the widely expected $150 by Wall Street analysts may only be a matter of time **

Huge divergence in the trend of chip stocks in the US stock market

The Philadelphia Semiconductor Index (SOX), known as the "global semiconductor stock barometer," plummeted last week due to ASML's unexpectedly poor performance. After ASML dramatically released financial reports well below expectations on Tuesday, it rarelly dropped by 5.3%. However, after Taiwan Semiconductor's performance announcement on Thursday, the index resumed a strong upward trend. Nevertheless, in the entire US stock market, the core force driving the repeated highs of US stocks - the chip sector, has seen a huge divergence in the performance of individual stocks.

Among them, semiconductor equipment companies have significantly underperformed the Philadelphia Semiconductor Index. Semiconductor equipment giants such as ASML, KLA, and Lam Research were rarely included in the "leading decliners list," while chip companies closely related to AI such as Taiwan Semiconductor, Nvidia, and Broadcom saw a significant increase in stock prices.

Semiconductor equipment companies led the decline in the chip sector of the US stock market last week

Semiconductor equipment companies provide chip manufacturing equipment, focusing on core chip manufacturing processes such as lithography, etching, thin film deposition, multilayer interconnection, and thermal management. Therefore, compared to chip companies like Taiwan Semiconductor, Nvidia, AMD, and Broadcom that directly benefit greatly from the global AI layout frenzy, semiconductor equipment companies focusing on heavy assets still heavily rely on the overall chip industry demand rather than just AI-related chip demand. If PC, smartphones, and other end-side AI large models become popular globally, it may drive the entire chip industry demand into prosperity, but there are no clear signs of this happening yet. ASML's latest financial results highlight the sluggish overall demand in the chip industry, causing the stock prices of these companies to collectively weaken.

"We expect this divergence to continue, as assuming that all of this is driven by AI demand is entirely correct." said Ryuta Makino, a senior analyst from Gabelli Funds. He expects that this significant divergence path will still exist at least until 2025.

The financial data of chip companies is often seen as a barometer of the global economy, as chips are crucial for manufacturing a wide range of electronic products from high-performance servers in data centers to smartphones, electric cars, and dishwashers. Companies providing semiconductor equipment for manufacturing these chips are at the forefront of the industry.

Before chip manufacturers like Taiwan Semiconductor and Samsung start outsourcing chip production, it takes time to build, install, and test large chip manufacturing machines. Therefore, semiconductor equipment companies like ASML and KLA have an unusually long-term view of customer demand. Currently, they are issuing warning signals for all chip areas outside of artificial intelligence. For example, due to the continuous increase in customer inventory, the demand for chips in electric cars and industrial end markets is still declining, and the demand for chips in the broad sense of personal computers and smartphones is still in a sluggish state since 2022, as the demand for AI PCs and AI smartphones has not expanded significantly. **

In addition, Intel, which holds an important position in the chip manufacturing field, is cutting costs and delaying the construction of new chip factories as the company is struggling to cope with a significant decline in sales and increased losses. This month, another chip manufacturing giant, Samsung, even directly apologized to investors, stating that the progress of its foundry business and delays in shipping HBM memory chips have disappointed investors.

Investors will closely watch the financial report of Texas Instruments this week. The company is set to release its financial report after the market closes on Tuesday Eastern Time, and its analog chip products are widely used by a diverse customer base. Texas Instruments can be said to have the widest customer base and the largest product range among chip manufacturers, so the company's performance and performance outlook data can serve as one of the predictive indicators for various industry demands. Texas Instruments is the world's largest analog chip manufacturer, and its products perform simple yet crucial functions, such as converting power to different voltages in electronic devices. In addition, Texas Instruments' analog chips have played an indispensable role in various key functional modules and systems in electric vehicles in recent years, including power management, battery management, sensor interfaces, audio and video processing, and motor control.

Overall, companies like ASML and KLA seem to be facing a more challenging path than expected in the semiconductor equipment sector, even though stocks of many semiconductor equipment companies such as KLA and Applied Materials hit historic highs earlier this year. Some Wall Street traders have not chosen to wait for signals of a recovery in overall chip industry demand and have already started selling stocks of semiconductor equipment companies like ASML.

ASML just experienced its worst week since early September, with its US stock price plummeting by 14%. The stock price of the largest semiconductor equipment company in the United States, Applied Materials, fell by 9.1%, while KLA Corporation's decline exceeded 12%.

"We are clearly more cautious on other semiconductor equipment names," analyst CJ Muse from Cantor Fitzgerald wrote in a research report. "We used to believe that longer-standing participants like ASML would perform potentially better. Clearly, our assumption was wrong." Following ASML's announcement of significantly lower-than-expected performance, the analyst stated that he expects the stock prices of semiconductor equipment companies to further decline.

AI Spending Continues to Surge, NVIDIA May Long Dominate the Title of "Biggest Winner of the AI Boom"

For "hot AI chip stocks" such as Taiwan Semiconductor, NVIDIA, Broadcom, AMD, and Micron Technology, which have been continuously attracting investments since 2023, the situation of their stock prices is quite different. These companies may continue to benefit from the continuous massive spending of major tech companies in artificial intelligence development.

Wall Street statistics show that Microsoft, Google's parent company Alphabet, Amazon, and Facebook's parent company Meta had capital expenditures exceeding $50 billion in the second quarter, with most of it used to purchase data center AI chips, such as NVIDIA's H100/H200/Blackwell series AI GPUs and Broadcom's AI ASIC. **Some tech giants have publicly stated that they plan to invest larger sums of money in the coming quarters to purchase AI GPUs, data center Ethernet chips, and other artificial intelligence infrastructure to expand or build new data centers According to the latest industry chain order information of the NVIDIA Blackwell GB200 chip released by the well-known technology industry chain analyst from TF International Securities, Microsoft is currently the world's largest GB200 customer, with Q4 orders surging 3-4 times, exceeding the total orders of all other cloud service providers.

In a recent report, the analyst mentioned that the capacity expansion of the Blackwell AI GPU is expected to start in the early fourth quarter of this year, with shipments expected to be between 150,000 to 200,000 units in Q4. It is projected that the shipment volume in Q1 2025 will significantly increase by 200% to 250%, reaching 500,000 to 550,000 units. This means that NVIDIA may achieve its sales target of one million AI server systems in just a few quarters. NVIDIA's founder and CEO Jensen Huang revealed in a recent interview that the Blackwell architecture AI GPU has been fully put into production and the demand is extremely "crazy."

According to the latest forecast data from Citigroup, one of Wall Street's financial giants, by 2025, the data center-related capital expenditures of the four largest tech giants in the United States are expected to increase by at least 40% year-on-year. These massive capital expenditures are largely related to generative artificial intelligence, indicating that the demand for computing power for AI applications such as ChatGPT remains significant. Citigroup stated that this means that the tech giants are expected to continue to significantly expand their spending on data centers beyond the already strong 2024 spending levels. The institution predicts that this trend will continue to provide very significant positive catalysts for the AI GPU dominator NVIDIA and data center interconnect (DCI) technology providers.

The four tech giants referred to in Citigroup's research report are the global cloud computing giants Amazon, Google, Microsoft, along with the social media company Facebook and its subsidiary Instagram. In this latest research report, Citigroup predicts that by 2025, the data center capital expenditures of these four tech giants will increase by 40% to 50% year-on-year. The substantial increase in spending by tech giants in data centers is expected to drive the stock prices of data center networking technology giants such as NVIDIA and Arista Networks, which are considered "shovel sellers" in the global artificial intelligence field, to continue to be favored by international funds.

The Citigroup analysis team stated in the latest research report that NVIDIA's absolute leading position in terms of total cost of ownership (TCO) and return on investment (ROI) in the AI infrastructure field, including server GPUs and NVIDIA's complete server hardware system, is emphasized as a core consideration for data center operators. They emphasize the higher level of performance running various applications (including AI training/inference applications) on NVIDIA's hardware and CUDA collaborative acceleration software platform. The Citigroup analysis team also emphasized that the adoption of artificial intelligence (AI) is still in the early to mid-stage, especially with the hot AI application drive of "AI agents" on the enterprise side, the demand for AI applications will surge, leading to a strong demand for computing resources, stimulating significant expansion and construction of data centers by tech giants Chief Investment Officer Solita Marcelli from UBS Global Wealth Management Americas predicts that by 2025, the total sales of chip companies closely related to artificial intelligence will jump from this year's $168 billion to $245 billion. Marcelli advises clients to increase their positions in stocks of AI-related chip manufacturers after ASML announces its performance.

"We continue to see strong growth prospects for AI chips and closely monitor management's guidance on future demand," she wrote in a research report last week.

According to Wall Street analysts, the primary beneficiary of all data center spending is NVIDIA, which continues to dominate the AI chip market broadly. With CEO Jensen Huang ensuring that its new Blackwell chip is in full production and seeing strong demand from customers, the stock hit a historical high last week, surpassing its previous peak in June. NVIDIA's stock price has risen by over 175% in 2024, currently with a market value close to $3.4 trillion, just a step away from surpassing Apple once again to become the "world's most valuable company."

TIPRANKS compiled data shows that 42 Wall Street analysts have an average target price expectation for NVIDIA within 12 months of $153.86, implying a potential upside of nearly 10%. The Wall Street banking giant Bank of America recently reiterated its "buy" rating on NVIDIA and significantly raised its target price from $165 to $190, higher than the general target price on Wall Street. The Bank of America analysis team also raised NVIDIA's earnings per share expectations for the fiscal year 2025 from $2.81 to $2.87; for the fiscal year 2026, from $3.90 to $4.47; and for the fiscal year 2027, from $4.72 to $5.67.

In the view of Wall Street analysts, other chip companies that can benefit from the surge in AI spending include Taiwan Semiconductor, Broadcom, Arm, Micron Technology, and AMD, especially as AMD is trying to weaken NVIDIA's absolute control over the AI chip market.

AMD's heavyweight AI accelerator MI300X has significant advantages in memory bandwidth and capacity compared to NVIDIA's Hopper architecture AI GPU, especially suitable for demanding generative AI model training and inference tasks with high requirements for AI parallel computing power Recently, some analysts believe that if AMD can continue to improve its NVIDIA CUDA competitor - the AMD ROCm software ecosystem, and accelerate its support for mainstream AI developer environments, it may further erode NVIDIA's share in the data center AI GPU market.

However, even "hot AI chip stocks" seem unable to avoid the negative impact of sluggish non-AI demand. For example, Broadcom, whose customized AI chips and Ethernet chips are mainly used in major data centers worldwide, saw its stock price plummet last month due to disappointing performance in some non-AI related aspects, and Broadcom's latest performance forecast basically indicates that the growth rate of its non-AI business is much slower than the market's expected recovery pace.

"Ultimately, there will be value scenarios for non-AI chip companies, especially when the economic recovery signals a rebound in demand," said Tim Ghriskey, Senior Portfolio Strategist at Ingalls & Snyder. "However, it's a matter of timing. AI will be the market focus in the long term."