SoftBank begs Gulf for $100B again as $20B problem looms

Nine years ago, Masayoshi Son walked out of the Gulf with nearly $100 billion in commitments for the first Vision Fund — the biggest tech fund ever assembled, bankrolled largely by Saudi Arabia's Public Investment Fund and Abu Dhabi's Mubadala. It gave the world ByteDance. It also gave the world WeWork at a $47 billion valuation. Now Son is back in the Gulf, and according to the Financial Times, he's asking for another $100 billion — this time while his company's shares are sliding, his biggest bet just saw its revenue numbers disputed by $20 billion, and the region he wants the money from is in the eighth month of a war.

The report landed Friday, October 9, and it says Son has held discussions in recent weeks with senior figures, including in the United Arab Emirates, about a potential fundraising for a new AI investment fund. SoftBank declined to comment. Reuters said it could not independently verify the report. We checked the wire copies against the FT's original claim: every outlet carries the same caveat — this is early-stage courtship, not a closed deal. But the ambition is unmistakable, and the timing is, frankly, astonishing.

Photo: Masayoshi Son at the SoftBank Mobile Summit in 2008 (file photo). The SoftBank chief is now courting Gulf investors for a new $100 billion AI fund, per the FT. Credit: nobihaya, via Wikimedia Commons, CC BY 2.0.

What the FT says Son wants

Per the FT's reporting, relayed by Reuters and the Japan Times, the new fund would acquire companies outright and then improve their operations using AI and other advanced technologies — including machines from SoftBank's own robotics arm, Roze. That's a meaningful shift in the playbook. Vision Fund 1 and Vision Fund 2 were primarily about writing enormous cheques into startups and riding them to an exit. This would be closer to a private-equity roll-up: buy real businesses, wire AI into their operations, and harvest the efficiency gains.

The Japan Times, which carried the richest version of the story, adds the only detail about who exactly Son has been talking to: "senior figures, including in the United Arab Emirates." No names, no term sheet, no fund size beyond "up to" $100 billion. And the paper notes, almost in passing, that Son is hunting for capital in a region "grappling with the threat of economic contraction from the Iran war, now entering its eighth month" — a sentence that would have been unthinkable in the launch-era press releases of 2017.

Why Son needs other people's money

Here's the part nobody at SoftBank will say out loud: the balance sheet is creaking. SoftBank has now put roughly $64.6 billion into OpenAI — the final $10 billion tranche of a $30 billion commitment closed on October 1, giving it a 13% stake. To fund that mammoth bet, the company raised $11.1 billion in September in what Reuters called the largest high-yield corporate bond sale ever recorded globally. Junk debt. At eye-watering coupons. A separate ¥1 trillion retail bond issue added another ~$6.3 billion of borrowing.

This is where the FT's $100 billion ask gets its real meaning. As Finimize put it in the sharpest analysis of the story, the Gulf pitch would shift SoftBank's playbook "from mainly using its own balance sheet — and sometimes pricey debt — toward managing a pool of outside money, more like a traditional private-equity structure where outside backers supply much of the cash." Translation: Son wants the Gulf's money so he can stop paying junk-bond rates for his AI dreams. If the money doesn't materialize, his deal pace stays chained to borrowing conditions — and borrowing is getting expensive.

The $20 billion problem hanging over everything

The fundraising news broke on the worst possible day. On Thursday, the FT reported — citing documents shared with OpenAI's financial backers — that OpenAI's annualized revenue stood at close to $50 billion at the end of September, roughly $20 billion less than the $70 billion figure investors had been led to believe from earlier reports. Asian markets flinched on Friday. SoftBank's shares dropped as much as 7.3% in Tokyo, per the Japan Times. Nvidia fell about 3%. The Nasdaq slid more than 1%.

The discrepancy, per Barron's, appears to be an accounting argument rather than a business collapse: the earlier $70 billion number reportedly counted gross revenue including cloud-partner revenue, while the $50 billion figure is closer to net. Chris Weston, head of research at Pepperstone, told reporters "the discrepancy may reflect differences in revenue measurement rather than an outright deterioration in OpenAI's underlying business" — but added that the headlines triggered "a meaningful one-day repositioning" in a market that has priced AI companies on relentlessly rising revenue curves.

Stephen Innes, managing partner at SPI Asset Management, put the mood more colourfully, writing that "Then OpenAI handed the market a number that made traders look at the other side of the ledger." For most of this boom, he noted, the market was happy to follow the money going in. Friday was the day it started asking about the money coming out.

The IPO that isn't happening

There's a second timing problem. SoftBank's entire OpenAI position — $64.6 billion of it — has no exit in sight. Sam Altman told Fortune in an interview released October 3: "Given everything happening with safety, right now would be an ill-advised moment to go public... I would say not 2026, yeah. We got a lot of stuff to do." Instead of an IPO, OpenAI is raising at least $30 billion more in a pre-IPO round targeting a ~$1.4 trillion valuation, per Bloomberg.

So picture Son's position: tens of billions sunk into a company that keeps raising more, at valuations that keep climbing, with no public listing to crystallize the gains — while his own shares wobble on revenue fears. The $100 billion Gulf ask starts to look less like ambition and more like arithmetic. He needs fresh, outside capital because the inside capital is spoken for.

2017 all over again — but the world has changed

The historical echo is the whole story here, and it's the angle nobody has fully mined. In 2017, Son raised nearly $100 billion for Vision Fund 1 in an era of free money. The Gulf's sovereign funds were awash in oil revenue and hunting for a post-oil future. The record since is genuinely mixed: ByteDance was a generational winner; WeWork was a colossal, public flameout that became shorthand for everything wrong with the era.

2026 is a different planet. Interest rates are punishing — SoftBank's own bond sale proves it. The IPO window that was supposed to recycle Vision Fund capital into new bets keeps slamming shut. And the Gulf investors Son is courting are managing portfolios through a regional war in its eighth month, with oil at $91–104 a barrel and their attention split between defense spending and economic contraction risk. The wallets are the same. Everything else is harder.

Our take: this is an admission, not just an ambition

Strip away the $100 billion headline and the FT's anonymous sourcing, and the story reads as a quiet confession: the era of Son funding the AI revolution off his own balance sheet is over. The $11.1 billion junk-bond sale was the proof; the Gulf roadshow is the consequence. That is actually the healthy version of events — outside money with outside discipline, a real private-equity structure instead of one man's conviction writing cheques at 2 a.m.

But here's the harder question the cheerleaders aren't asking: if the plan is to buy ordinary companies and "improve their operations using AI," Son is no longer competing with venture capitalists. He's competing with every private-equity firm on earth that has discovered the same pitch — and many of them can borrow more cheaply than he can. Vision Fund 1 worked because Son saw mobile and software scale before the consensus did. This time, there is no informational edge. Everyone knows AI improves operations. The edge, if there is one, is just the size of the cheque. And even Son now has to ask the Gulf for that.

Frequently asked questions

Is the $100 billion fund confirmed? No. The Financial Times reported it citing people familiar with the matter; Reuters said it could not independently verify the report, and SoftBank declined to comment. Treat it as early talks.

What would the fund do? Per the FT, it would acquire companies and improve their operations using AI and advanced technologies, including robots from SoftBank's Roze arm — a buy-and-modernize model closer to private equity than venture capital.

How much has SoftBank already put into OpenAI? About $64.6 billion cumulatively, including the final $10 billion tranche that closed October 1, 2026 — giving it a roughly 13% stake.

Why did SoftBank's shares fall 7.3% on Friday? The drop was driven by the FT's separate report that OpenAI's annualized revenue is ~$50 billion rather than the ~$70 billion previously flagged — not by the Gulf fundraising report itself.

Sources: Financial Times reporting via Reuters and the Japan Times (Oct 9, 2026); Reuters wire via CNA and The Standard; Finimize analysis; TipRanks; Barron's on the revenue-accounting dispute; Economic Times and Malay Mail market coverage; Fortune interview with Sam Altman (Oct 3, 2026); Bloomberg on OpenAI's $1.4 trillion bridge round.

Related: SoftBank Wires Final $10B, Locking In Son's Colossal $64.6 Billion Grip on OpenAI — how the October 1 tranche completed the biggest private AI bet ever made. SoftBank Is Raising $11 Billion in Debt to Keep Feeding OpenAI — the record junk-bond sale that showed the balance sheet straining. OPENAI REVENUE: The $20B Gap That Rattled Tech Stocks — the accounting dispute behind Friday's sell-off.

By Daniel Rao
Daniel covers AI policy, platform security, and the geopolitics of technology for TechPulse USA.

The market's instant verdict: SoftBank shares slid 7.3%

The market did not wait for confirmation. SoftBank shares dropped as much as 7.3% in Tokyo on Friday (Oct 9), weighed down by concerns over the pace of OpenAI's revenue growth, according to Reuters reporting. The scale of the bet explains the jitters: the company has invested almost $65 billion in OpenAI ahead of the lab's public debut, while grappling with growing borrowing costs after raising approximately $11.1 billion through a high-yield bond sale in September. And the leverage keeps building: SoftBank has increased a margin loan secured on its Arm shares to about $25 billion, with a $40 billion bridge loan coming due in March 2027. Son has held the early talks with senior figures including in the United Arab Emirates, and the proposed fund would buy companies and improve their operations using AI and machines — including from SoftBank's robotics arm Roze, which is expected to play a key role, the FT reported. A SoftBank representative declined to comment; there is no guarantee the discussions result in a deal. (The wider debt binge: the AI debt machine.)

What we still don't know

Three honest caveats: there is no SoftBank confirmation — a representative declined to comment; there is no guarantee the talks become a deal; and no anchor investors have been named yet. Reuters could not independently verify the FT's account. (OpenAI's delayed IPO is the overhang behind all of this.)