(Image generated by Baidu Wenxin Yige AI)
The new AI craze seems to be showing signs of "going out."
Titanium Media App reported on April 20 that as of the close of the U.S. stock market on Friday, the share price of Supermicro Computer (NASDAQ: SMCI), the AI server leader of "ten baggers a year", plummeted by more than 20% overnight, and the latest closed down 23%, hitting a new low in more than two months.
Meanwhile, Nvidia (NASDAQ: NVDA) shares fell 10%, the lowest day since March 2020, and wiped out more than $210 billion (about 1.52 trillion yuan) in market value overnight.
In the past week, Nvidia's market value has evaporated by nearly $300 billion - about the same as one AMD ($237 billion), two Intel ($145.585 billion), and 1.8 CATL ($839.337 billion).
On the news side, as a core server supplier of NVIDIA, Supermicro Computer announced on April 19 that it would announce its results for the third fiscal quarter of 2024 on the 30th of this month, but did not give more details of the financial report. This "vanishing notice" is considered to be unusual, causing fears that the AI bubble is about to burst, which has caused many AI stocks in the United States to suffer heavy losses.
It's clear that the new AI boom is shifting dramatically from the perspective of capital markets.
AI concept stocks suffered a "pullback", and Supermicro and Nvidia plummeted by more than 23% and 10% respectively in a single day
U.S. AI concept stocks ushered in a tragic "Black Friday".
On April 19, after Supermicro announced the timing of its third-quarter results for fiscal year 2024, Wells Fargo Securities issued a research report pointing out that it did not give a positive advance statement and did not disclose important AI data, which was seen as a negative signal.
Other analysts pointed out that Supermicro's failure to release a preliminary revenue report as usual could mean that its quarterly results were weaker than expected.
Affected by this news, Dow Jones market data showed that at the close of trading on Friday, Supermicrocomputer's stock price plummeted 23.14%, down more than 40% from its all-time high in March, while Nvidia's stock price plummeted 10%, the biggest one-day decline since March 16, 2020. The stock price plummeted nearly $85, setting a new record for the largest one-day decline in history.
At the same time, Nvidia's rival AMD fell 5.4%, chip design company Arm fell nearly 17%, wafer foundry leader TSMC fell more than 3%, chip manufacturers Broadcom and Marvell fell 4.3% and 4.8% respectively, and memory chip supplier Micron fell 4.6%.
Avinash Gorakshkar, head of research at Profitmart Securities, said that the main reason for Nvidia's plunge was that the hardware partner only announced the date of the earnings announcement and did not disclose preliminary results. Analysts expect Nvidia's stock price to remain sideways to negative until Supermicro's results are announced.
It's not just a one-day drop in stock prices. In the past week, the stock price of the "Seven Sisters of the US Stock Market" continued to fall heavily, and the total market value evaporated by $950 billion, the worst week in history.
In terms of stock prices, Tesla was the top loser, falling more than 14% this week. However, in terms of the amount of market value evaporation, Apple, Microsoft, and Nvidia contributed the most, and the market value of these three companies far exceeded that of Tesla.
Nvidia was the tech giant that lost the most in market value this week, with a loss of nearly $300 billion.
Jordan Klein, a resident analyst at Mizuho Securities, said that "the entire industry has retreated" in the chip field, and the speed of the retracement has been getting faster and faster in the past week or so.
It is worth noting that Tesla's market value shrank by $76 billion to about $468 billion this week, ranking all the way down to 14th among listed companies in the United States, while Tesla's market value was overtaken by Walmart on Thursday and ExxonMobil on Friday, from January 2023 to last week, Tesla's market value ranking has been in front of these two companies, while Apple's market value has evaporated by $178 billion this week, Microsoft's market value has shrunk by $169 billion, Amazon's market value has evaporated by $118 billion, Google's parent company Alphabet's market value has shrunk by $41 billion, and Meta's market value has shrunk by $68 billion。
According to Bloomberg, the Seven Sisters fell below their 50-day moving average for the first time since October last year, with a combined market value of $950 billion, far exceeding the record of $872 billion set in January 2022.
Investors think there are too many bubbles, but TSMC says AI chip revenue has doubled
At present, the AI industry has entered a controversial point in time.
On the one hand, investors have expressed concern about the "bursting of the bubble" of the AI boom, making investment more cautious, and on the other hand, AI applications have exploded, and investors have also shifted from semiconductors and AI-based models to software and scenarios.
Recently, the Institute for Human-Centered Artificial Intelligence (HAI) at Stanford University, which Li Feifei co-leads, recently reported that global investment in AI has declined for the second year in a row over the past year.
Among them, AI-related corporate investment (M&A and acquisition) decreased by 31.2% from $117.16 billion in 2022 to $80.61 billion in 2023, while private investment (i.e., venture capital investment in start-ups) fell from $103.4 billion to $95.99 billion.
Taking into account minority deals and public offerings, total AI investment fell to $189.2 billion last year in 2023, a 20% decrease compared to 2022.
Gartner analyst John-David Lovelock said that with first-tier players such as Anthropic and OpenAI dominating, the scope of AI investment is "spreading out".
"The amount of billions of dollars in investment has slowed and is almost over, and hot money is pouring in a new direction – AI applications. "Large models require significant investment, but the market is now more influenced by tech companies that will leverage existing AI products, services, and offerings to build new products." ”
Touring Capital联合创始人Samir Kumar认为，繁荣时期不会持续下去。
"We will soon evaluate whether generative AI can deliver the promised efficiency gains at scale and drive revenue growth through AI-integrated products and services. If these expected milestones are not achieved and are still in the experimental phase, the revenue from the 'experimental run rate' may not translate into sustainable annual recurring revenue. Samir Kumar said.
Seth Rosenberg, a partner at Greylock, believes that interest in funding a "large number of new players" in the AI space is inherently small. In the early stages of the cycle, investment in underlying models was very capital-intensive, compared to the low capital required for AI applications and agents, which may have contributed to the decline in absolute dollar funding.
Umesh Padval, managing director at Thomvest Ventures, attributed the overall reduction in AI investment to lower-than-expected growth. He said that the initial enthusiasm has given way to reality – that AI faces some technical challenges, some go-to-market challenges, and it may take years to solve and fully overcome.
"The slowdown in AI investment reflects the recognition that we are still exploring the early stages of AI technology development and its application across industries. While the long-term market potential remains enormous, the initial enthusiasm has been tempered by the complexity and challenges of rolling out AI technology in real-world applications...... This indicates that the investment environment is more mature and sensitive. Umesh Padval said.
In addition, the recent financial reports of semiconductor stocks have also begun to "sound the alarm" to the market, which seems to have also "woken up" to investors in AI concept stocks.
For example, semiconductor equipment manufacturer ASML (ASML) reported poor financial results, sales performance fell more than expected, the company's stock price has fallen for three consecutive days, TSMC also lowered the outlook for the overall semiconductor market for the whole year at the legal conference this week, causing many semiconductor concept stocks around the world to fall.
However, TSMC also said that the AI server chip market still faces new opportunities in the future.
TSMC President Wei Zhejia also highlighted in the earnings call on April 18 that AI technology is developing rapidly, models are becoming more and more complex, and more powerful semiconductor hardware support is needed.
"Whatever path you take, you need to use state-of-the-art semiconductor process technology. He expects that in 2024, TSMC's AI server chip-related revenue will at least double, accounting for 11%~13% of total revenue, and in the next five years, it will maintain a compound annual growth rate of 50%, accounting for more than 20% of the company's total revenue by 2028.
Wei Zhejia pointed out that server AI chips will become the strongest driving force for the growth of the company's HPC business, and will become the largest source of revenue growth for the company in the next few years. The company plans to mass-produce the 2nm process in the fourth quarter of 2025.
He also said that almost all AI chip innovators are working with TSMC, and customers have shown high interest and participation in TSMC's 2nm process, and it is expected that the number of new tape-outs in the first two years of mass production will be higher than the previous 3nm and 5nm processes.
TSMC expects second-quarter revenue to be between $19.6 billion and $20.4 billion. Based on the median of the above range, TSMC's second-quarter revenue is expected to increase by 6% quarter-on-quarter and 27.6% year-on-year. TSMC also expects its overall business to be stronger in the second half of 2024 than in the first half.
Counterpoint, a market research firm, pointed out that TSMC's 5nm capacity utilization has reached full capacity, thanks to strong demand from Nvidia's AI GPUs. Continued strong demand for advanced chips, especially for AI, is a positive sign for both the short and long term.
(This article was first published on Titanium Media App, author | Lin Zhijia, editor | Hu Runfeng)