Masayoshi Son has never been accused of thinking small. The SoftBank founder once invested $300 million in Alibaba before the company was worth anything. He also poured billions into WeWork, which was… less successful. Now he wants to borrow $40 billion, a record for SoftBank, to double down on OpenAI. And if you’re thinking ‘that’s a lot of borrowed money,’ welcome to the AI economy in 2026.
The Numbers Are Absurd (In a Very Real Way)
Here’s what happened in the last two weeks of February: OpenAI closed a $110 billion funding round backed by Amazon, Nvidia, and SoftBank. That single round pushed the ChatGPT maker’s valuation to roughly $730 billion, making it one of the most valuable private companies in history.
SoftBank was already in for a major chunk. They held about 11% of OpenAI at the end of last year. But Son apparently looked at that number and thought: not enough.
According to Bloomberg, SoftBank is now seeking a loan of up to $40 billion, denominated in dollars, with a 12-month tenor. JPMorgan Chase is reportedly one of four banks underwriting the deal. The financing would be SoftBank’s largest dollar-denominated borrowing ever.
To put that in perspective: $40 billion is roughly the GDP of Paraguay. Or about 120 Lamborghini factories. Or, if you prefer a simpler comparison: it is a lot.
Why Is Son Doing This?
SoftBank had a rough patch. The Vision Fund era of 2019-2022 saw some spectacular misfires, WeWork being the most memorable. Son pulled back, reorganized, and spent a couple of years convincing the world he’d learned his lesson about overleveraged bets on visions of the future.
Then AI happened. And Masayoshi Son, never a man for cautious optimism, apparently decided this was the moment to go all-in again.
OpenAI is reportedly preparing for an eventual IPO that could value the company at around $1 trillion. If that happens, an 11% (or larger) stake becomes an extraordinary return on investment. The bet makes sense, in a ‘go big or go home’ kind of logic that Son has always been comfortable with.
The bridge loan structure is telling too. Twelve months is short. This reads less like a long-term financing play and more like: we need the capital now, we’ll sort the rest out when OpenAI goes public.
The Elephant in the Room
Not everyone is cheering. A $40 billion loan to fund a single AI position is the kind of move that makes risk managers sweat through their spreadsheets.
OpenAI is extraordinary by most measures. Its products are used by hundreds of millions of people. Revenue is growing fast. But it burns money at a scale that’s almost difficult to grasp: building data centers, training frontier models, hiring the best engineers on earth. The path to profitability exists, but it is not short.
And there is competition. Google has Gemini. Meta is open-sourcing everything. China’s AI ecosystem is accelerating hard, as shown by Beijing’s new five-year plan, which mentions ‘AI’ more than 50 times and treats it as national economic infrastructure. The market OpenAI dominates today could look very different in 24 months.
SoftBank is betting that none of that matters, that OpenAI’s lead is durable enough to justify infrastructure-scale financial exposure. That might be right. It might not be.
What This Tells Us About the AI Race in 2026
The larger signal here is about what kind of race AI has become. This is no longer a software story. It is a capital-intensive infrastructure story, similar in scale to telecom buildouts or semiconductor fabs in previous decades.
The players who win the next phase of AI are not necessarily the ones who ship the best models. They are the ones who control the compute, the distribution, and the financing. SoftBank, with its position in OpenAI and its telecom infrastructure, is trying to own all three vectors at once.
Whether that strategy works depends on a lot of variables nobody can fully predict. But you cannot say Son is not committed.
The Pudgy Cat Take
Borrowing $40 billion to buy more of a company valued at $730 billion, which itself raised $110 billion just days ago… this is either the most rational bet in tech history or a spectacular slow-motion catastrophe in the making.
The truth is probably somewhere in between. OpenAI is real, its products work, and its trajectory is pointing up. But the assumptions baked into a $730 billion private valuation are enormous, and $40 billion in debt is a very expensive way to discover you were wrong.
Masayoshi Son has been right about transformative tech before. He has also been spectacularly, expensively wrong. The most honest take: nobody really knows how this ends. We just know it is going to be dramatic either way.