After Masayoshi Son regained the title of Japan's richest man, he liquidated his NVIDIA holdings. Does this signal the bursting of the AI bubble?
Masayoshi Son Is Set to “Miss Out on” NVIDIA Again
A year ago, when Masayoshi Son of SoftBank and Jensen Huang of NVIDIA embraced and shed tears, lamenting the premature sale of NVIDIA shares, most people believed that NVIDIA's value was unrivaled. If given a second chance, Son would probably regret not buying more and holding the shares for a shorter period.
However, Masayoshi Son always comes up with unexpected moves. He has once again liquidated all his NVIDIA holdings. SoftBank Group announced that it had sold all its NVIDIA shares, cashing out approximately $5.8 billion (equivalent to about 41.5 billion yuan). Subsequently, NVIDIA's stock price dropped in the evening, and its market value evaporated by hundreds of billions of dollars overnight. What a “masterstroke”! Is there a more worthy investment than NVIDIA?
Source: Xueqiu
Surprisingly, there is. It's OpenAI, the core enterprise of this AI revolution. Without it, the story of NVIDIA wouldn't exist. But is it too early to bet on OpenAI's explosion in the application field? We'll analyze this later.
There are certain similarities between Masayoshi Son's two liquidations of NVIDIA shares. In Son's dictionary, there is no concept of diversified hedging investment. Instead, he focuses on heavily investing in leading concepts, going all - in or going home.
Previously, there was the failed case of WeWork. In 2020, its valuation was $70 billion. After the failure of its IPO, the bubble burst, resulting in an annual loss of tens of billions. Later, the valuations of the unlisted technology companies in which SoftBank had heavy investments plummeted in 2022, with an annual loss of up to $32 billion, setting a historical record. In addition, investments in the South Korean e - commerce company Coupang and the “Amazon of the food industry” Zume also resulted in significant losses.
However, since the first half of this year, by betting on NVIDIA, acquiring Arm and ABB's robotics business, and building an AI full - industry chain layout, Masayoshi Son, the investor who was “mocked by the whole network,” has regained the title of the richest man in Japan. For a while, he became the most inspiring and powerful investor on the Internet.
Source: Internet
But Masayoshi Son will only disappoint those who praise him. Recently, SoftBank sold its NVIDIA shares, cashing out approximately $5.83 billion, sold its T - Mobile shares, cashing out approximately $9.2 billion, and mortgaged its Arm shares. By using its stake in the chip - design company as collateral, it expanded its guaranteed loan amount to $20 billion.
It is well - known that SoftBank liquidated all its NVIDIA holdings for $3.6 billion in 2019 and then repurchased them in 2020 until this latest sell - off. The previous sell - off has become a cautionary tale in the investment circle. If SoftBank Group had retained those original shares, their value would now exceed $150 billion.
Many people focus only on Masayoshi Son's failure to hold onto NVIDIA shares. However, the more important reason is that at that time, SoftBank's focus was on autonomous driving and AI investments. In 2019, Son clearly did not consider NVIDIA as the core stock in the AI field. It is common for him to abandon mature layouts for other more promising investments.
Although later in 2020, he planned and facilitated the merger of ARM and NVIDIA, with a transaction price of up to $40 billion. Unfortunately, this attempt was not approved by the regulators, and Masayoshi Son missed out on NVIDIA once again.
Source: Internet
It is not difficult to see that Masayoshi Son's investments are more from a financial perspective, often starting from the potential performance and market space rather than the perspective of value investment. This was the case last time, and it is the same this time. OpenAI, with a valuation of only $500 billion, is obviously undervalued. Therefore, it is reasonable to shift the focus and heavily invest in OpenAI. Due to Son's investment style, he often either invests in the wrong company or invests too early or too late. So, which category does the heavy investment in OpenAI fall into?
Masayoshi Son Is All - In on OpenAI's IPO
Overall, currently, OpenAI seems to be in its early stage rather than the stage where application revenue can explode.
“Some people say I have DeepSeek anxiety and spend too much on AI,” Masayoshi Son has publicly stated more than once. He believes that in the next decade, AI will affect at least 5% to 10% of the global GDP. “If the return is in the range of $9 trillion to $18 trillion per year, why be stingy? The competition is getting fiercer. We must take the lead, be brave, and go all - in.”
When asked about the timing of selling NVIDIA shares during the earnings conference call, Yoshimitsu Goto, the Chief Financial Officer (CFO) of SoftBank Group, said that the company needed liquidity to fulfill its investment commitments to OpenAI. He stated, “This year, our investment in OpenAI is huge, exceeding $30 billion. Therefore, we have to liquidate some of our existing assets.”
Source: Internet
But is it really as simple as just changing the heavily - invested stock? Is it possible that SoftBank is bearish on NVIDIA's future valuation and worried that the AI bubble will burst more severely at the top? Currently, it seems more like a signal of “early escape.”
Coincidentally, Bridgewater Associates, the largest hedge fund on Wall Street, released its quarterly portfolio report as of the end of the third quarter this year on Friday. The report showed that Bridgewater significantly reduced its holdings of NVIDIA, the chip giant, by nearly two - thirds in the third quarter. It also cut more than half of its Alphabet shares, about 9.6% of its Amazon shares, and more than 35% of its Microsoft shares. In a report to investors, Karen Karniol - Tambour, the Chief Investment Officer of Bridgewater, and others warned that the current market stability is facing increasing risks.
When asked whether the liquidation of NVIDIA shares involved a valuation judgment, Yoshimitsu Goto declined to comment, only saying that adjusting the asset allocation is the fate of an investment company.
Of course, Masayoshi Son won't say that he is bearish on NVIDIA, but some busybodies have smelled something fishy.
Last week, SoftBank also announced a “1 - for - 4” stock split plan, which will take effect on January 1, 2026. Looking back at history, since the 21st century, every stock split by SoftBank has been accompanied by significant global market turmoil.
Stock split (1:3) in June 2000: The dot - com bubble burst.
Stock split (1:3) in January 2006: The U.S. housing bubble burst.
Stock split (1:2) in June 2019: The global market tumbled at the beginning of 2020 due to the COVID - 19 pandemic.
So, what kind of storm will this stock split bring? This is not an alarmist statement. The bond market seems to have smelled the danger. The interest rate of SoftBank's recently issued bonds has exceeded 8%, indicating that bond investors are demanding a higher risk premium. David Gibson said, “But the stock market seems to have ignored this potential, under - funded risk.”
SoftBank's trust in OpenAI may exceed that in any other company in the past. In the investment return from OpenAI, SoftBank has recorded a profit of $8 billion without actually investing real money, just based on a promise.
Wall Street analysts point out that there may be a gap of up to $54.5 billion between the total investment commitments made by SoftBank and the actual available funds, posing a risk of over - commitment.
In the latest earnings report, OpenAI was mentioned 100 times, while NVIDIA was only mentioned 14 times.
Source: Internet
Can SoftBank make a comeback by relying on OpenAI and be fearless of the bubble risk? Probably not.
An executive of OpenAI mentioned in a public event that “investment in computing power may need government guarantees.” This statement sent shockwaves through the market, causing a sharp decline in U.S. technology stocks. The market value of the six major technology giants evaporated by more than $400 billion overnight.
“The U.S. stock market is facing a serious risk of decline. There may be a sharp correction in the next 6 months to 2 years,” said Jamie Dimon, the CEO of JPMorgan Chase, a prominent figure on Wall Street. He believes that the asset price increase driven by AI is a “worrying issue, and many assets in the market seem to be entering the bubble zone.”
In October this year, Sam Altman, the CEO of OpenAI, also rarely stated that there is indeed a bubble in some parts of the AI field, and some “stupid” startups can easily obtain large amounts of funds.
Some institutions estimate that OpenAI is expected to lose more than $5 billion in 2025. If it fulfills its commitment of a $1.4 - trillion infrastructure investment, the total cash - burning scale in the next few years may exceed $100 billion. Even with the support of giants like Microsoft, the continuous huge cash consumption puts great pressure on the company. The market generally worries that under hardware bottlenecks such as a shortage of chips, the cost of AI training will only increase. Once investors' confidence is shaken, it is easy to trigger a chain reaction of the “burst of the technology bubble.”
Currently, OpenAI is unable to solve this funding problem and can only adopt a “circular transaction” model to support the market. For example, while Microsoft invests approximately $13 billion in OpenAI, it designates its Azure as OpenAI's exclusive cloud service provider.
Source: Internet
Another example is NVIDIA. This year, it agreed to invest up to $100 billion in OpenAI to jointly build large - scale data centers, and OpenAI promised to purchase millions of NVIDIA GPUs to fill these racks. Oracle, an early investor in OpenAI, also signed an agreement this year to provide cloud services worth $300 billion to OpenAI over the next five years.
Paul Krugman, the Nobel laureate in economics, compared this phenomenon to an “Ouroboros that devours itself.” He said, “On the surface, it is the income generated from sales, but in fact, it is just the same funds circulating among different companies.”
David Cahn, a partner at Sequoia Capital, put it more bluntly. He said that if a company can only survive in an environment of “infinite financing,” then there is something wrong with the company itself. In the current context, he is neither blindly chasing stories like “reaching $2 million in ARR in 10 days” or “serial financing with a three - fold increase in valuation” nor ignoring the risks. Excessive funds may make founders and teams think they have already won, but the real challenge is to survive after the momentum fades.
Can OpenAI successfully go public? Probably. But the biggest problem for OpenAI is not a lack of funds but a shortage of a huge amount of electricity.
By 2028, U.S. data centers alone will require 57 gigawatts of electricity, while utility companies can only provide a maximum of 21 gigawatts at that time, resulting in a huge gap of 36 gigawatts. In addition to electricity, there are also shortages in land, water resources, fiber - optic networks, and even the production capacity of key grid equipment such as transformers.
Source: Internet
These physical bottlenecks cannot be solved by financial means alone. Circular transactions create an illusion of “unlimited demand,” but they may hit a wall in the face of the reality of “limited supply.”
Conclusion
Historically, each round of technology bubbles has different manifestations, and the bursting of a bubble does not mean the failure of a technological revolution. A typical example is the dot - com bubble in the early 2000s. The huge investment losses did not equal the failure of the Internet revolution. Moreover, after a brief downturn, the Internet revolution quickly gained momentum.
Therefore, industry insiders deeply involved in AI often say that “the risk of not investing is much greater than the risk of investing.” No one wants to be left far behind by competitors in the context of an exponential revolution.
Companies that are likely to laugh last are probably not those in the “oil - to - chip” business. Those that use computing power to create products have a chance to come out on top in the face of the bubble.
References:
The $10 - trillion AI Bubble. Source: LatePost
SoftBank Liquidates NVIDIA Holdings, Cashing Out $5.8 Billion. Source: FORTUNE
Masayoshi Son Cashes Out and Bets on OpenAI. Source: Top News of Yicai