On November 11th, Beijing time, SoftBank Group sold off all of its Nvidia shares. The sale of shares, netting $5.83 billion, has sparked concerns in the market about a potential bursting of the AI bubble. But the sell-off is not over yet.
Bridgewater Associates, Wall Street's largest hedge fund, released its holdings report on Friday, showing that it significantly reduced its stake in chip giant Nvidia by nearly two-thirds. Bridgewater also reduced its holdings in Alphabet by more than half and its Amazon holdings by approximately 9.6%. Stocks, over 35% of Microsoft Stocks. In a report to investors, Bridgewater's chief investment officer, Karen Karniol-Tambour, and others warned that current market stability is facing increasing risks.
Furthermore, Citigroup's latest US stock holdings report shows that the institution also reduced its holdings in Nvidia , Microsoft , and Apple in the third quarter. And the four major US tech giants, including Amazon . However, Citigroup recently raised its target price for Nvidia from $210 to $220 per share.
In recent weeks, discussions about an AI bubble have intensified.
Some company executives have warned that the stock market may be heading for a downturn. For example, Jarek Kutylowski, CEO of AI company DeepL, said this week that many market valuations are quite exaggerated, and there are signs that a bubble is imminent. Hovhannes Avoyan, CEO of image editing tool company Picsart, said it's worrying to see many AI companies receiving extremely high valuations without much revenue.
In addition, Michael Burry, a well-known short seller who accurately predicted the 2008 US subprime mortgage crisis, expressed concerns this week about the AI infrastructure construction boom. He stated that major artificial intelligence... Infrastructure and cloud service providers underestimate the depreciation costs of chips, like Oracle. Such companies' profits may be significantly overestimated; he was also recently shorting Meta and Nvidia. Ben Harburg, managing partner of investment firm Novo Capital, stated that the investment amounts being discussed by large tech companies may be exaggerated, particularly regarding data centers. Construction may have become excessive.
Some argue that selling Nvidia stock does not necessarily mean a loss of confidence in the company, and that the timing of the sale may not be wise.
SoftBank's sale of its Nvidia shares is primarily to fund its bets on artificial intelligence projects. Market analysts believe SoftBank needs these funds for strategic moves, including the "Stargate" project with OpenAI and Oracle , which requires large-scale data center cluster construction. SoftBank also plans to invest $40 billion in OpenAI and is betting on the chip industry. Following its acquisition of Arm and planned $6.5 billion acquisition of chip design company Ampere, recent reports suggest SoftBank is considering acquiring Marvell.
SoftBank Group founder Masayoshi Son stated last year that he regretted selling his Nvidia shares in 2019. CJ Muse, senior managing director at investment bank Cantor Fitzgerald, also stated that Son's timing in the Nvidia stock transaction was not excellent, and the sale was primarily for resource allocation purposes, aiming to invest funds elsewhere.
On the contrary to the view that AI is overvalued, some practitioners in the technology industry are still optimistic about the long-term potential of AI.
David Risher, CEO of ride-hailing app Lyft, recently stated that we are currently in a financial bubble due to the transformative technology brought about by AI. No one wants to be left behind by the technological wave. Data centers and models have long lifespans. Although tech companies are taking some financial risks now, these investments will make life better in the future.
Whether the tech industry's investment in artificial intelligence can translate into substantial returns will be a key factor in determining whether the AI bubble will burst.
A recent report by venture capital firm Accel estimates that new AI data center capacity will reach 117 gigawatts by 2030, implying nearly $4 trillion in related capital expenditures over the next five years, requiring approximately $3.1 trillion in revenue to repay these expenditures. The firm believes that generating this revenue will require more powerful AI models, new AI services, and a revolution in agent applications within enterprises.
AI chip Industry giants remain optimistic about the expanding demand for AI and the prospects for data center construction. AMD CEO Lisa Su recently stated that the data center chip and system market is expected to expand to $1 trillion by 2030, and the company's annual revenue from data center chips will reach $100 billion within five years.
Nvidia will release its third-quarter earnings report next week, providing its latest performance. Nvidia's stock price exceeded $200 per share at the end of October, subsequently correcting somewhat, closing at $190.17 per share on November 14th, with a market capitalization of $4.62 trillion. Barclays Analysts say investors’ caution ahead of Nvidia’s earnings report is understandable, as the report may be crucial in determining whether the AI supercycle and related spending are still on track.
(Article source: CBN)
