After the sharp fall in early August, the stock price of American AI concept head companies rebounded sharply.
On August 5th, Nvidia's stock price was around 100 yuan per share, and on August 28th it closed at $128.3 per share. On August 5th, the stock prices of Microsoft, Google, Meta, and Apple fell below $400/share, $160/share, $480/share, and $210/share, respectively. On August 28th, they closed at $413.84/share, $164.68/share, $519.1/share, and $228.03/share, respectively.
The concept of artificial intelligence in the US stock market continues to be hot, but some experts are packing their bags and leaving one after another. Buffett's holdings of Apple Inc. stock have been sold off by half, and Berkshire Hathaway's cash reserves have reached a historic high. As a barometer of the global artificial intelligence industry, NVIDIA CEO Huang Renxun has continuously reduced his holdings in the past three months, cashing out approximately $520 million worth of NVIDIA stock.
David Roche, a senior US strategist, predicted that in 2025 there would be a bear market (a decline of 20%) in US stocks, one of the reasons being the artificial intelligence foam.
Is there a foam in AI?
Among the artificial intelligence concept stocks, Nvidia has the strongest stock price performance. From the beginning of this year to June 7th, the last trading day before the "one to ten" stock split, Nvidia's stock price rose by 144.12% this year, more than seven times higher than at the beginning of 2023. The peak of Nvidia's stock price occurred in mid June, with a stock price of $135.58 per share and a market value of $3.335 trillion. For the other six of the "seven sisters of technology stocks", the share price peak in the year also appeared from mid June to mid July, with Microsoft's share price exceeding $460, Apple's over $230, Google (Alphabet) - A's over $190, Meta's over $530, Amazon's over $200, Tesla's over $260.
From the perspective of static P/E ratio PE (LYR), in the past five fiscal years, Google PE (LYR) has decreased from 30.05 in fiscal year 2019 to 29.15 in fiscal year 2023. Amazon, on the other hand, experienced negative values due to losses in 2022 and returned to positive in fiscal year 2023. Apple PE (LYR) has risen from 16.61 to 26.82, Microsoft from 39.33 to 45.91, Meta from 26.47 to 39.21, AMD from 152.6 to 180.41, and Nvidia from 37.02 to 345.12. Among them, Nvidia's latest fiscal year PE (LYR) is the highest in the past five years. For companies such as AMD, Meta, Google, Apple, and Microsoft, whose PE (LYR) has fluctuated and risen, the latest fiscal year's PE (LYR) is not the highest in the past five years, indicating that these stocks are showing signs of overvaluation.
Qu Shaojie, Assistant General Manager of the International Business Department of Great Wall Fund and QDII Fund Manager of Great Wall Global New Energy Vehicles, believes that the overvaluation of top technology stocks is driven by market outlook. At the early high point, the general rule for US technology leaders is a P/E ratio of around 30 times. At the previous high point, the P/E ratio exceeded the average P/E ratio by 0.7 times the standard deviation, indicating a slight overestimation. After the recent correction, the P/E ratio has returned to around 25 times the average, which is 0.5 times the standard deviation, and is already relatively healthy, "said Qu Shaojie. He told reporters that the P/E ratio is an important valuation indicator to judge whether there is a foam. On the one hand, the P/E ratio is compared with the historical average. For example, the valuation of technology companies at this stage is slightly higher than the average in recent five years, but there is no obvious overestimation. More importantly, the price to earnings ratio needs to match the growth rate of the next year or years. The higher the expected growth rate in the future, the higher the price to earnings ratio may be.
From the logic of injecting funds into top technology stocks in the US stock market, Zhang Chi, Chairman of Beijing Xinding Rongsheng Capital Management Co., Ltd., stated that there may not necessarily be a necessary relationship between stock price fluctuations and operating data. The background of the rise in the stock price of the leading technology stocks in the US is the role of the US dollar in raising interest rates. A large part of the global capital is piled up in the "seven sisters of science and technology", and the funds have banded together to stir up the stock price and make mutual offers with the concept of artificial intelligence. The market value and trading volume of these seven stocks account for a large proportion of the US stock market, so a sharp decline in any one stock will cause a catastrophic drop in the US stock market.
Zhang Chi believes that the foam of the "Seven sisters" in the US stock market is very serious. The US stock market has been rising continuously for 10 years, and without concept speculation, it is difficult to support. The rise in stock prices can be understood as a result of the accumulation of funds in the market, which lacks some self convincing evidence. At present, some funds in the market have been withdrawing from top technology stocks in the US, but as long as the "last moment" is not reached, many funds will not withdraw in large quantities. It is expected that this "last moment" will come soon.
Zhang Lu, a well-known technology investor in Silicon Valley and founding partner of Fusion Fund, has a different view. "Some companies' stock prices have not really reached a high multiple, but the seven companies we often talk about have a relatively fast growth rate, showing a very obvious Matthew effect. This is also because on the one hand, people are very excited about artificial intelligence, and on the other hand, they are not very clear about how artificial intelligence can quickly bring value in the short term." She believes that a relatively safe and balanced approach is to invest in some already stable giants, believing that these giants have both resources and channels, and the ability to replicate and promote artificial intelligence solutions.
Dismantling the "AI Content" of US Tech Leaders
At the end of 2022, ChatGPT emerged, triggering a wave of generative AI. At present, the so-called artificial intelligence is mainly based on generative AI, and is used as the core for research and development, production, sales, and various applications. From this background, the "AI content" of the "seven sisters of technology stocks" in the US stock market is quite different, and can be roughly divided into the following three categories according to the "AI content".
The first category is chip and cloud service companies that provide AI infrastructure.
Among chip companies, Nvidia's performance has been significantly catalyzed by AI in the past year or two. In the annual report ending January 28, 2024, Nvidia's revenue increased by 126% year-on-year and net profit increased by 581% year-on-year. In the quarter ending on that date, Nvidia's data center business revenue closely related to AI accounted for 83%. In the latest quarter ending on April 28th this year, Nvidia's revenue increased by 262% year-on-year, and the proportion of data center business to total revenue increased to 87%.
In contrast, AMD's association with AI may be weaker. In the second quarter of fiscal year 2024, AMD's revenue and net profit increased by 9% and 19% respectively year-on-year. The data center business unit, which is closely related to AI, saw the most rapid revenue growth in the quarter, with a year-on-year increase of 115%. This business accounted for approximately 48% of revenue. But compared to Nvidia, which mainly sells GPUs, AMD's data center division still receives a considerable portion of its revenue from CPUs.
Microsoft, Google, and Amazon can be classified as cloud vendor arrays. From the perspective of cloud business proportion, Microsoft's association with AI may be the strongest, while Google and Amazon's cloud business revenue proportion is relatively low. In addition to cloud services, these vendors actually have other "main businesses". Although some of their businesses are also related to AI, it is still uncertain how much revenue growth AI has driven.
Among them, Alphabet, the parent company of Google, had a revenue of $84.742 billion in the second quarter of fiscal year 2024, with cloud business revenue of $10.347 billion and advertising business revenue of $64.616 billion, accounting for 12.2% and 76.3% of the revenue, respectively. Microsoft's quarterly revenue as of the end of June this year was $64.7 billion, with its AI related intelligent cloud division generating $28.5 billion in revenue, accounting for 44% of the total revenue. Amazon's net sales for the second quarter of 2024 were $148 billion, of which the cloud platform AWS had a net sales of $26.3 billion, accounting for 17.8% of the revenue. Taking Google as an example to analyze the incremental revenue brought by AI to the company, if compared with the second quarter of the 2022 fiscal year before the launch of ChatGPT, assuming that both Google's cloud business and advertising business revenue incremental in the second quarter of this year come from AI, then the revenue incremental brought by AI is about 12.3 billion US dollars, accounting for about 14% of the revenue in the second quarter of this year.
It is worth noting that Amazon's main business is still e-commerce, with a net sales revenue of $55.39 billion in the latest quarter, significantly higher than its cloud business. Microsoft and Google's other businesses are somewhat related to AI, such as Google's forward-looking product development and venture capital "other bets" business related to AI, but the revenue scale of this business has not yet become a trend. Microsoft's productivity and business process businesses, as well as personal computing businesses, are also somewhat related to AI, but the driving force behind revenue remains to be observed.
The second category is companies that apply AI, which has a significant pulling effect on Internet advertising business.
As a social media giant, Meta's main source of revenue comes from advertising. In the second quarter of 2024, Meta's revenue was $39.071 billion, with its App series generating $38.718 billion in revenue, including $38.329 billion in advertising revenue, a year-on-year increase of 21.7%. Meta also has a revenue of $353 million from Reality Labs, which includes VR and AR products related to AI landing applications, but the revenue scale of this business has not yet reached a peak.
The seventh sisters of technology shares also has a third kind of company, namely Apple, which is relatively special. In the fiscal quarter ending June 29th, Apple's revenue was $85.8 billion, of which $61.564 billion was from hardware sales such as iPhone, and $24.213 billion was from hardware related services. AI has not yet brought enough direct changes to Apple's financial data, and the industry is mainly observing whether AI functions will drive hardware sales, and expects that Apple Intelligence intelligent systems may shift towards charging in the future.
Overall, AI mainly drives the hardware sales of top chip manufacturers in the US stock market, bringing some incremental growth to cloud computing businesses and improving advertising and other businesses for cloud manufacturers. However, the role of AI in promoting intelligent hardware sales still needs to be demonstrated. Apart from significantly boosting the performance of AI chip companies like Nvidia, the impact of AI on some non cloud businesses of cloud vendors is relatively limited.
In Qu Shaojie's view, artificial intelligence companies can be divided into three different roles: chip companies, model companies, and application companies. The development of AI can be divided into three different stages: the stage of AI big model development, the stage of AI diffusion into vertical fields, and the stage of AI application explosion. It can be inferred that the development of AI is still in its first stage, and chip manufacturers have performed well, making them the companies with the largest stock price increases
The AI content is a dimension to observe whether the "seven sisters of technology stocks" in the US stock market have a foam. The other dimension is to see whether the changes in capital expenditure, income and profit of these seven leading enterprises are synchronized. If capital expenditure is soaring at a high speed, income and profit are not keeping up. The greater the gap between them, the deeper the foam will be.
In terms of capital expenditures, Google's parent company Alphabet spent $13.2 billion on purchasing technology infrastructure in the second quarter of fiscal year 2024, which is expected to be higher than last year. This year, the company's quarterly capital expenditures are expected to be no less than $12 billion; Meta expects capital expenditures to reach $37 billion to $40 billion in 2024, with a significant increase in capital expenditures next year; Microsoft's latest quarterly capital expenditure reached $19.652 billion; Apple stated that the company has been investing in AI and machine learning, and capital expenditures are expected to increase year by year. Amazon's capital expenditure in the first half of this year was $30.5 billion, and it is expected that this expenditure will be even higher in the second half of the year. In the first half of this year, capital expenditures of Microsoft, Google, Amazon, and Meta increased by 50%, totaling over $100 billion, reaching a historic high.
However, in terms of performance, the growth rate of giants that increase capital expenditures is not synchronized. Meta's net profit in the second quarter increased by 73% year-on-year; Google's parent company Alphabet's net profit for the second quarter increased by 29% year-on-year; In the latest fiscal quarter ending at the end of June this year, Microsoft's net profit increased by 10% year-on-year. Microsoft's AI related cloud business generated $28.5 billion in revenue for this quarter, lower than market expectations of $28.7 billion; Amazon's net profit in the second quarter increased by 101% year-on-year, with AI related AWS business revenue of $26.281 billion, higher than market expectations of $26 billion. However, due to increased fixed asset expenses, AWS business profit margin decreased by 2 percentage points to 36%.
Intel has even tasted the bitter fruit of increased capital expenditures without timely performance boost. Intel's net loss for the second quarter of fiscal year 2024 was $1.6 billion, and management stated that the lower than expected profitability for the quarter was partly due to the company's decision to accelerate production of Ultra AI CPUs. Another tech giant seems to have accepted the fact that capital expenditures cannot be timely reflected in returns. Alphabet executives have stated after the financial report conference that these projects, including land acquisition, construction of data centers, and financial leasing, may take 15 years or even longer to realize.
From the business structure and capital expenditures of US tech giants, it is likely that AI's reshaping of these companies' businesses is just beginning, and increasing capital expenditures at present is based on more forward-looking planning. Whether capital expenditures can be converted into revenue in the future will be a key factor in whether these giants can maintain high stock prices.
Will the foam burst
Can the share prices of the seven sisters of technology stocks keep rising? Some market analysts believe that it is not only important to look at the operating performance of these technology companies, but also to consider market sentiment and global capital flows. If we consider market sentiment and capital flow, this wave of overvalued technology stocks may be unavoidable.
Zhang Chi told reporters that behind the concentration of US stocks is the mentality of capital seeking stability, and the risk of funds piling up in small stocks is higher. And there is also a possibility of investors' investment in large projects dispersing in a panic. If the US dollar cuts interest rates and global funds leave the United States, it is expected to accelerate the decline in the stock prices of these tech giants.
After experiencing a recent pullback in US tech stock prices, some market participants predict that market volatility may be greater in the second half of the year. Qu Shaojie judged that the market's expectation of interest rate reduction in September continued to rise. If the United States entered a new cycle of interest rate reduction, the U.S. bond market would be affected. Some funds might withdraw from the U.S. bond market and invest in the equity market, some might invest in growth stocks, and some might seek high dividend assets. The market volatility may be greater in the second half of the year than in the first half. The uncertain factors are that the US presidential election has not yet ended, and the impact of industrial policies on the market after the new president takes office is still uncertain. In addition, the Federal Reserve's interest rate cuts have yet to be implemented, and the convening of a Federal Reserve meeting may disturb the capital market.
Zhang Chi believes that the seven sisters of technology stocks have experienced a sharp decline recently. In addition to the unsatisfactory employment data in the United States, it is also related to the market's awareness that AI stocks should not be hyped too high. The market is gradually realizing that the revenue that AI can monetize is limited, and its main application scenarios are still chatting and drawing. It is not clear what kind of applications AI can have in the industrial field. After more than a year of development, AI has not changed much in the manufacturing industry, has little effect on increasing industrial scale, and has limited support for GDP growth.
Looking back, we still need people to solve the problem. In order to manage the foam, we still need to maintain technological progress through continuous R&D investment. First, only when the speed of innovation keeps up with the expected hype will the foam not burst. Secondly, by smoothly transitioning from an overheated economy back to equilibrium and achieving a 'soft landing', market sentiment can be stabilized. Any drastic change in market sentiment may lead to a chain reaction foam or collapse, because NASDAQ has reached a critical threshold after several rounds of technology speculation. Again, it is necessary to maintain sufficient liquidity levels by moderately raising interest rates. If one or more of these three conditions are not met, the current NASDAQ foam may burst.
Zhang Lu believes that when evaluating investment projects, more and more attention is being paid to whether they can establish business connections with large enterprises and provide value. Market validation is regarded as one of the measurement standards, with a focus on the quality and sustainability of customers who actually purchase products. The correlation between labor costs and operational efficiency is weakening, and the application of large models may lead to the automation of certain tasks. The key is to have a deep understanding of the business system of the target market. What we value more is actually the expandable space, resources, and channels in the original business of large companies, rather than their so-called technological reserves.