SoftBank is seeking a loan of up to $40 billion — its largest dollar-denominated borrowing ever — primarily to finance its investment in OpenAI. The deal would cement SoftBank’s position as one of the biggest backers of frontier AI and signal that the AI race has entered infrastructure-scale economics.
According to Bloomberg, the bridge loan would have a tenor of about 12 months, with four lenders including JPMorgan Chase underwriting the facility. The financing comes days after OpenAI completed a $110 billion funding round backed by Amazon, Nvidia, and SoftBank that pushed the company’s valuation to $730 billion.
The Numbers Are Staggering
Let’s put $40 billion in context. That’s larger than the GDP of over 100 countries. It’s more than the entire annual revenue of companies like Intel or Goldman Sachs. SoftBank is borrowing at a scale typically reserved for sovereign nations or mega-mergers.
The loan would help SoftBank increase its stake in OpenAI beyond its current roughly 11% holding. At a $730 billion valuation, that stake alone is worth roughly $80 billion — making it the single most valuable position in SoftBank’s portfolio by a massive margin.
Reuters reported last year that OpenAI had begun laying the groundwork for an IPO that could value the company at around $1 trillion. If SoftBank can increase its stake before that happens, the return on this loan could be enormous.
AI Bets Now Look Like Infrastructure Plays
The scale of this financing tells you something important about where AI economics have landed. These aren’t software investments anymore. They’re infrastructure bets — closer in scale and structure to what you’d see in telecom buildouts, energy projects, or semiconductor fabrication.
OpenAI’s $110 billion round wasn’t about building a chat app. It’s about funding compute infrastructure, training runs that cost hundreds of millions of dollars, and building the foundational layer of AI that every other company will eventually run on.
SoftBank is betting that whoever controls that layer controls the next era of technology. And it’s willing to leverage its entire balance sheet to be at the table.
Masayoshi Son’s Redemption Arc
SoftBank’s founder Masayoshi Son has spent the last few years recovering from the Vision Fund’s high-profile failures — WeWork, Wirecard, and a string of overvalued startups that cratered during the 2022 downturn.
OpenAI represents the opposite end of the spectrum: a company with real revenue (reportedly $13 billion annualized), massive distribution (ChatGPT), and a product that’s reshaping how every industry works.
If OpenAI’s IPO hits $1 trillion, SoftBank’s stake could be worth well over $100 billion. The $40 billion loan suddenly looks like the smartest trade of Son’s career.
If it doesn’t work out, it’s the most expensive one.
What This Means for Markets
The loan highlights a broader trend: AI is now large enough to reshape corporate balance sheets and global lending markets. When a single company’s investment thesis requires $40 billion in bridge financing, you’re no longer in startup territory. You’re in systemic territory.
The four banks underwriting this deal — including JPMorgan — are making their own bet that AI valuations will hold. If they’re wrong, $40 billion in exposure becomes a problem that ripples through the financial system.
For now, the bet is that OpenAI’s trajectory — from $1 billion in revenue in 2023 to $13 billion in 2025 to a potential IPO at $1 trillion — is unstoppable. That’s a lot of faith in a company that still hasn’t proven it can be sustainably profitable.
The Concentration Risk
SoftBank is concentrating an extraordinary amount of capital into a single company, in a single sector, at what may be peak valuations. If the AI bubble deflates — or if a competitor (DeepSeek, Anthropic, Google) erodes OpenAI’s moat — the downside is catastrophic.
But Son has always been a conviction investor. He backed Alibaba when no one else would. He’s betting that OpenAI is this generation’s Alibaba moment. The $40 billion loan is him putting everything on the table.
Sources: Bloomberg, Reuters, TechStartups