SoftBank sold Nvidia: is this an example for the retail investor?
SoftBank had already exited Nvidia in 2019: it was one of the worst investment decisions ever, as the value of the stake sold then would now exceed $150 billion
This week it became known that Japanese investment conglomerate SoftBank completely got rid of its stake in Nvidia in October 2025, selling 32.1 million shares for $5.83 billion. Oninvest asked Egor Tolmachev, senior analyst at Freedom Broker Capital Markets Research, whether this deal should be considered a benchmark for retail investors.
Sell everything for OpenAI
SoftBank has conducted one of the largest asset sales in recent years - the group sold stakes in T-Mobile, Deutsche Telekom and sold its entire stake in Nvidia - to free up capital to invest in OpenAI. According to the company's CFO Yoshimitsu Goto, this is a major strategic realignment of SoftBank's investment portfolio.
There's a lot of investment in OpenAI this year - over $30 billion to be invested. So we really need to sell some of our existing portfolio to use that money for financing. It doesn't make a lot of sense that this happened in October, and it has nothing to do with NVIDIA itself
In March 2025, SoftBank entered into an agreement to invest up to $40 billion in OpenAI, of which $10 billion has already been contributed in capital in April, with the remaining $22.5 billion scheduled to be invested in December 2025.And although the decision on the deal was made by the Japanese investor a long time ago, the market took it as an illustration of the transition to a new stage of the AI technological revolution. Masayoshi Son, SoftBank's founder, is changing "shovels for mines": while Nvidia produces AI tools, OpenAI directly manages value-creating AI models.
The thin spots of the AI boom
SoftBank's sale of its stake in Nvidia coincided with the growing debate about a possible bubble in the AI market, which only added fuel to the fire. Michael Burry, who predicted the 2008 crisis, opened short positions against Nvidia and Palantir in the last quarter. He said the companies overestimate the real useful life of chips and hardware, and talk about the rapid depreciation of older chips.
Understating depreciation by extending the life of assets artificially increases profits - one of the most common types of fraud in the modern era
Veteran investor Mark Mobius has warned of a possible 30-40% correction in AI stocks due to inflated valuations.
The past quarter really highlighted this weak side of AI's investment appeal - high valuation levels of key AI companies with unobvious returns on investment in the new technology. The disappointment of the quarter was Meta's investor call, when management, responding to a question about excessive data center construction, made it clear that the company could always sell this computing capacity. That had the opposite effect - the stock was down nearly 9% on the postmarket that day. The growing amount of AI infrastructure also caused many to think again about rising depreciation and amortization (D&A) charges.
OpenAI, Nvidia and Oracle form a strategic triangle in which investment, technology and capital circulate. AI startups already account for 53% of all venture deals, and among the largest investors are often the same bigtechs - Microsoft, Google, Amazon, Meta. The situation when invested funds are used to purchase goods and services of the investor company is a common business practice in itself, but the current scale and number of such deals raises the degree of investors' concerns.
SoftBank paid for the turnaround: what it means for investors
Interestingly, the immediate market reaction to the news of the sale was negative for SoftBank's own stock, not Nvidia's.
After the media reported the sale of the stake in Nvdivia, SoftBank's shares in Tokyo were falling 10% in the moment, later correcting down to 3.5%.
All this happened despite the fact that the company reported a net profit of $16.2 billion for the second quarter - almost five times more than analysts' expectations. Nvidia's shares fell by 3% at the end of the trading session on November 11. For investors, the SoftBank deal itself does not change much, and in a sense, it removes some of the concerns about investments in AI, because the proceeds from the sale will be used, among other things, to finance the construction of data centers and the purchase of equipment.
Nevertheless, investors saw the move as SoftBank exiting a stable, profitable asset for a risky bet on OpenAI. Notably, this is SoftBank's second full exit from Nvidia. In 2019, the company sold its stake in Nvidia for about $3.6 billion, which was one of the worst investment decisions, as they would now be valued at over $150 billion. The company returned to investing in Nvidia again as early as 2020.
This was not the Japanese investor's only miscalculation, and after catastrophic losses of $14.3 billion on his WeWork investment, Son temporarily shifted to a more cautious strategy. By the end of 2023, he had already announced a return to "attack mode", focusing on the topic of AI. This illustrates well Masayoshi Son's characteristic style - aggressive, large-scale bets on future technologies with a willingness to radically change positions.
What a retail investor should do
According to Vanda Research, by the end of 2024, Nvidia has become the second position after Tesla in the portfolios of "typical retail investors" in the U.S., accounting for more than 10% of their assets, up from 5.5% at the beginning of 2024. In January 2025, following news of DeepSeek's breakthrough model, retail investors bought a record $562 million worth of Nvidia stock in a single day.
For the market, the SoftBank deal serves as a reminder of the need to critically evaluate valuations and risks, especially with the unprecedented concentration of capital in the AI sector. Retail investors should focus on diversification, fundamentals and the long term, avoiding the euphoria that often precedes bubbles. Ultimately, history shows that technological revolutions do change the world - but the way to do so is through cycles of enthusiasm, corrections and consolidation. The AI era is no different from the old days: investors who can distinguish long-term fundamental trends from speculative noise will benefit. Currently, 60 of the 66 analysts following Nvidia stock have a Buy recommendation. We see no reason to sell Nvidia stock and maintain a Buy recommendation with a 12-month target price of $210 - which implies a 10.4% upside from the Nov. 14 closing price.
This article was AI-translated and verified by a human editor
