Masayoshi Son sells complete stake in chip giant, despite Anthropic’s clearer path to profitability
November 12, 2025 — In one of the most dramatic strategic shifts of the AI investment cycle, SoftBank has liquidated its entire Nvidia position—more than 32 million shares worth approximately $5.8 billion—to fund a massive bet on OpenAI.
The move represents a fundamental repositioning by CEO Masayoshi Son, who is pivoting away from AI infrastructure toward what the company views as the next wave of value creation: AI applications and platforms that consumers and businesses actually use.
The Investment Strategy
SoftBank’s aggressive capital reallocation includes:
- Complete Nvidia exit: $5.8 billion from selling entire stake
- T-Mobile stock sale: Over $9 billion liquidated
- ARM margin loan: Additional leverage secured on existing holdings
- OpenAI commitment: Participation in secondary share sale that closed last month, plus leading a $40 billion primary funding round expected to close soon
SoftBank emphasized this isn’t a bearish call on Nvidia’s future or valuation. Rather, the firm believes the “interface layer” of AI—platforms, agents, and user-facing tools—will generate superior returns compared to infrastructure investments.
A Costly Precedent
This marks the second time SoftBank has exited Nvidia, having previously sold its position in 2019. That earlier decision proved extraordinarily expensive: had SoftBank held those shares, they would be worth $210 billion today, given Nvidia’s 5,000% surge since 2019.
Even for Japan’s richest man, that represents one of the most expensive missed opportunities in investment history.
The Profitability Challenge
The timing of SoftBank’s all-in OpenAI bet raises questions, as newly surfaced financial documents reveal stark differences between OpenAI and its primary competitor, Anthropic.
According to analysis of internal financial projections reported by The Wall Street Journal:
OpenAI’s Path:
- Consumer-heavy revenue mix (70% consumer, 30% enterprise)
- Expected losses of $74 billion in 2028 alone
- Burning 14 times more cash than Anthropic
- Profitability not expected until 2030
- Q3 2025 cash burn: $12 billion
- Projects “hundreds of billions” in revenue by 2030
Anthropic’s Path:
- Enterprise-focused strategy (85% business customers)
- Significantly faster trajectory to profitability
- Lower infrastructure costs
- More sustainable business model
The Consumer Product Dilemma
OpenAI’s challenges stem largely from its consumer-facing products. ChatGPT and the recently launched Sora video generation tool drive massive user engagement but remain “deeply unprofitable” due to extraordinary infrastructure requirements.
The company has committed over $1.4 trillion to power its consumer-facing operations—a staggering capital allocation that exceeds the market capitalization of most American companies.
Bernstein analysts estimate OpenAI burned through $12 billion in Q3 alone, based on patterns in Microsoft’s latest quarterly earnings.
The Bull Case
Despite mounting losses, OpenAI CEO Sam Altman projects the company will achieve revenue “on the scale of hundreds of billions of dollars by 2030″—a sharp upward revision from previous guidance.
The implicit argument: short-term losses represent necessary investment in market leadership, user acquisition, and technological advancement that will ultimately generate returns at unprecedented scale.
Market Implications
SoftBank’s move reflects a broader debate in AI investing: whether value will accrue primarily to infrastructure providers (like Nvidia) or to application-layer companies that monetize AI through direct customer relationships.
The firm’s willingness to deploy every available liquidity lever—equity sales, margin loans, and stake liquidations—signals conviction that OpenAI represents the winning side of that debate.
However, the company’s path to profitability remains uncertain, with projected losses that would exceed the annual GDP of many nations. Whether investors will maintain patience through nearly $100 billion in cumulative losses before 2030 remains the critical question.
As one analyst noted: “A lot of companies make billions. They try to make profit on those billions, not lose money. But it’s early. We’ll see how it shakes out.”
Analysis based on the financial news coverage and reporting by The Wall Street Journal on AI company financial projections.
