Deja vu? SoftBank Sold Nvidia Stakes, 2019 Exit Before Epic rally

Japan's SoftBank Group reported in its earnings that it completely divested its Nvidia holdings in October, selling 32.1 million shares and realizing $5.83 billion. SoftBank CFO Yoshimitsu Goto stated, "We wish to have ample investment opportunities while maintaining financial strength. The share sale is part of our asset monetization strategy."

Notably, SoftBank purchased $4 billion in Nvidia stock in 2017, then fully exited its position in January 2019. From a stock trend perspective, SoftBank clearly sold at a low point in 2019.

Multiple analysts indicated that SoftBank's move does not reflect a bearish view on Nvidia but aims to stockpile "ammunition" for major investments this quarter.

"In our view, this should not be seen as a cautious or negative stance on Nvidia. In Q4 2025, SoftBank requires at least $30.5 billion for investments, exceeding its total from the previous two years. This includes $22.5 billion directed to OpenAI and $6.5 billion to Ampere. The amount 'exceeds the sum of its investments over the prior two years in a single quarter,'" said Rolf Bulk, equity analyst at New Street Research.

SoftBank's CFO noted that Q2 profit doubling was thanks to its investment in OpenAI last September. This also explains why SoftBank is now significantly increasing its bets on OpenAI.

Morningstar analyst Dan Baker added that the move does not signal a shift in SoftBank's stance toward Nvidia but rather a need to recycle capital for investments in other AI sectors.

Moreover, even after divesting its Nvidia shares, SoftBank's business interests remain closely tied to Nvidia. It is involved in multiple AI projects reliant on Nvidia technology, including the $500 billion "Stargate" project for large-scale data center construction.

In a research report dated November 10, top investment bank Citi maintained its "Buy" rating on Nvidia stock and raised its price target from $210 to $220. Citi bets that Nvidia's upcoming earnings will again far exceed expectations, citing the shipment of 6 million GPUs disclosed at Nvidia's GTC conference in Washington. The primary contradiction in the AI chip market remains supply shortage, not lack of demand.