SoftBank Raises Record $11.1 Billion in High-Yield Bonds to Fund OpenAI and AI Ambitions

SoftBank Group has raised a record $11.1 billion in high-yield bonds to finance its expanding investment in OpenAI and other artificial-intelligence bets, using expensive debt to accelerate one of the world’s largest corporate commitments to the technology.

The Japanese technology investor sold dollar and euro denominated senior notes in what Reuters, citing LSEG data, described as the world’s largest high-yield corporate bond sale on record. The fundraising will help SoftBank make a $10 billion payment for the third and final tranche of its $30 billion follow-on investment in OpenAI, due to close Oct. 1, while also supporting general corporate purposes.

The transaction highlights both the scale of Masayoshi Son’s AI ambitions and the financial risk attached to them. SoftBank is not merely investing from accumulated cash or selling small parts of its portfolio. It is raising debt at interest rates that approach 10% to deepen exposure to a single, fast-growing but still commercially unproven AI ecosystem centered on OpenAI.

The sale attracted investors despite the cost, demonstrating that markets remain willing to finance bold AI strategies. At the same time, the yields demanded by buyers show that creditors expect to be compensated for the uncertainty.

A record bond sale

SoftBank’s $11.1 billion financing package consists of $10 billion in U.S.-dollar senior notes and €1 billion, or roughly $1.1 billion, in euro-denominated notes.

The U.S.-dollar offering was divided into three tranches:

Currency	Amount	Maturity	Interest rate
U.S. dollars	$1 billion	3½ years	8.625%
U.S. dollars	$4.5 billion	5½ years	9.25%
U.S. dollars	$4.5 billion	7½ years	9.75%
Euros	€500 million	4 years	7.125%
Euros	€500 million	6 years	8.0%

The sale surpassed the previous global record for a high-yield corporate bond issuance: French cable operator Numericable Group’s $10.9 billion sale in 2014, according to LSEG data cited by Reuters.

High-yield bonds, often called “junk bonds,” are issued by companies whose debt is rated below investment grade. The label does not mean the bonds are worthless. It means rating agencies and investors view the risk of default as higher than for investment-grade debt, so companies must typically offer higher interest rates to attract capital.

SoftBank has long operated as a speculative-grade borrower, but the yields on this offering show how dramatically financing conditions have changed. In June 2021, the company issued $7.3 billion of dollar- and euro-denominated senior bonds with yields ranging from 2.125% to 5.25%. This week’s dollar tranches carry rates from 8.625% to 9.75%.

That gap reflects higher global borrowing costs, but it also reveals investor caution about SoftBank’s increasingly concentrated AI exposure, its complex balance sheet and the scale of capital needed to sustain its OpenAI strategy.

The OpenAI commitment

The immediate purpose of the bond sale is clear: SoftBank needs cash for a major OpenAI payment.

According to the offering term sheet, the proceeds will fund a $10 billion payment that represents the third tranche of SoftBank’s $30 billion follow-on investment in OpenAI. That installment is expected to close Oct. 1.

SoftBank’s total commitment to the maker of ChatGPT now stands at $64.6 billion. Once the latest transaction is completed, SoftBank is expected to hold roughly 13% of OpenAI.reuters+1

Those figures place SoftBank at the center of OpenAI’s financial structure and make its future increasingly tied to the company’s ability to turn extraordinary user interest in generative AI into durable revenue.

OpenAI has become one of the most closely watched private technology companies in the world. Its products have helped popularize generative AI for consumers and businesses, while its models are used for writing, programming, research, customer support and software development. But the business also demands immense spending on specialized chips, data centers, electricity, engineering talent and safety research.

That creates an unusual financing challenge. A company can have rapidly growing revenue and global attention while still requiring huge amounts of outside capital to build and operate the computing infrastructure needed for next-generation AI. SoftBank is betting that OpenAI’s growth will eventually justify the cost.

A debt-funded wager on AI

The bond sale is not SoftBank’s only recent funding action.

Earlier this month, SoftBank issued ¥1 trillion, or about $6.3 billion, in bonds aimed at retail investors. Reuters reported that the company has also sold assets and taken loans secured against its holdings in chip designer Arm and OpenAI to help fund its commitments.

SoftBank has issued $14.6 billion in high-yield bonds so far in 2026, accounting for 63.4% of the Asia-Pacific and Japan high-yield corporate bond market, excluding Australia and New Zealand, according to Reuters.

The figures show the extent to which SoftBank has become a major driver of speculative-grade debt issuance in the region. Its borrowing strategy is helping transform the bond market into a financing channel for AI infrastructure and AI-company investments.

Masayoshi Son has long built SoftBank around large, concentrated bets on technological change. His record includes successes, including the company’s early investment in Alibaba, and setbacks, including investments whose valuations fell sharply after the easy-money era ended. The current AI campaign follows the same basic philosophy: secure a dominant position in a potentially transformative technology before its long-term winners are fully established.

The difference is scale. SoftBank’s OpenAI commitments alone total $64.6 billion. Financing that commitment with high-yield debt raises the possibility that the company’s balance sheet could be exposed if enthusiasm for AI fades, if OpenAI’s valuation falls or if capital markets become less accommodating.

Investors demand a premium

SoftBank’s ability to place the bonds demonstrates investor demand. Reports ahead of the final pricing said the offering drew more than $20 billion of interest, well above the size of the deal.

But strong demand should not be confused with low risk. Investors were attracted in part because SoftBank offered unusually high yields.

The 7½-year dollar notes pay 9.75%, the highest coupon SoftBank has offered on dollar debt, according to Reuters. The company’s 5½-year notes pay 9.25%, and even the shortest dollar tranche carries an 8.625% coupon.

For SoftBank, those rates create a substantial future interest burden. The company will pay hundreds of millions of dollars annually in interest on the new debt before repaying any principal. That can be manageable if AI investments increase in value and generate returns. It can become difficult if valuations decline, earnings weaken or refinancing markets tighten.

The deal therefore contains two messages from the market:

• Investors are willing to back SoftBank’s AI thesis.
• They require significant compensation for taking the risk.

That is a more nuanced verdict than either unqualified confidence or outright skepticism. Bond buyers are not necessarily agreeing that AI investments will succeed. They are judging that the potential return, at the offered yield, is sufficient to justify the risk.

Why high-yield financing matters

SoftBank’s transaction matters beyond the company itself because it illustrates a broader question facing the AI industry: Who pays for the infrastructure race?

The most advanced AI systems require massive capital investment. Companies need high-end chips, data-center space, networking equipment, power contracts, cooling systems and specialized engineering talent. Building this infrastructure can cost tens of billions of dollars before revenue catches up.

Large technology companies such as Microsoft, Alphabet, Amazon and Meta can finance much of that spending through cash flow generated by established advertising, cloud-computing and software businesses. OpenAI, by contrast, relies more heavily on partnerships and outside financing as it grows.

SoftBank’s debt issuance helps bridge that gap. It brings public-bond-market capital into an AI financing ecosystem that has previously depended heavily on venture capital, private equity, strategic investment and private credit.

The approach may accelerate AI development. It also amplifies financial risk because debt must be repaid regardless of whether an investment succeeds. Equity investors can lose value if a company performs poorly, but bondholders still expect interest and principal payments according to the terms of the debt.

That distinction is critical. SoftBank is using debt to buy exposure to a technology whose long-term economics remain unsettled. Generative AI has demonstrated enormous demand, but the path to sustained profitability remains unclear because operating advanced models can be costly and competition is intense.

Arm and the wider AI portfolio

SoftBank’s AI ambitions extend beyond OpenAI.

The company retains a major stake in Arm, the British chip designer whose semiconductor architecture is used widely in smartphones and increasingly in data-center and AI applications. SoftBank has used financing linked to its Arm and OpenAI holdings to support its broader investment strategy.

Arm gives SoftBank an important position in the foundational infrastructure of computing. OpenAI gives it exposure to the consumer and enterprise application layer. Together, those stakes form part of Son’s vision of SoftBank as a central investor in the AI economy, spanning chips, models, robotics, data centers and potentially autonomous systems.

The opportunity is substantial. If AI becomes as economically important as Son believes, early control of meaningful stakes in critical platforms could produce exceptional returns.

The risk is equally substantial. AI valuations may be based on expectations that take years to justify. If companies overbuild data centers, face regulatory constraints, struggle to monetize consumer products or encounter more effective competitors, the returns may not support the amount of capital flowing into the sector.

SoftBank’s strategy leaves relatively little room for a prolonged downturn. The more it borrows against AI-linked assets, the more vulnerable it becomes to shifts in market sentiment toward those assets.

What could go right

For SoftBank, the optimistic scenario is clear.

OpenAI continues to expand enterprise and consumer revenue, develops products that justify its valuation and becomes a dominant platform for AI services. Arm benefits from broader demand for energy-efficient processing in AI devices and data centers. SoftBank’s investments appreciate, allowing it to service its debt, refinance at more attractive rates and potentially sell portions of its holdings at a profit.

In that scenario, the bond sale will look like a bold but rational way to secure ownership in a transformational technology before it becomes even more expensive.

The company could also benefit if AI demand produces a sustained investment cycle. Data centers, semiconductors, robotics and enterprise software may all grow alongside frontier models. SoftBank’s willingness to make large commitments now could provide it with strategic access and influence in a market where scale increasingly matters.

What could go wrong

The risks are not hypothetical.

OpenAI’s valuation and SoftBank’s financial position could suffer if AI revenue fails to grow fast enough to cover operating and infrastructure costs. Competition from Google, Anthropic, Meta, Microsoft, Alibaba and other companies could reduce pricing power. Regulators could impose new restrictions. A slowdown in enterprise AI spending could reduce the demand that investors now expect.

SoftBank could also face refinancing risk. Debt issued at 8% to nearly 10% becomes expensive to maintain, and future borrowings could be even costlier if rates rise or credit markets turn against speculative-grade issuers.

The company’s reliance on stakes in Arm and OpenAI introduces market risk. A drop in either asset’s valuation could reduce the collateral value available for secured borrowing and undermine confidence in SoftBank’s balance sheet.

Reuters noted that SoftBank’s finances could become vulnerable if sentiment toward OpenAI and the AI sector weakens significantly.

That is the central caution embedded in the record sale. The bonds give SoftBank cash now. They do not guarantee that its AI investments will produce the returns needed to justify the debt.

A defining test for AI finance

SoftBank’s $11.1 billion bond sale is a landmark transaction because it turns the AI boom into a large-scale corporate-credit story.

Masayoshi Son is making a familiar argument in a new form: the companies that finance the next technology platform early will shape the future economy. Investors have given him the money to pursue that vision, but at a price that reflects the magnitude of the gamble.

The record issuance gives SoftBank the funding it needs for the next OpenAI payment and reinforces its place as one of AI’s most aggressive financial backers. It also raises the stakes. A $64.6 billion commitment to OpenAI, funded in part with costly debt, ties SoftBank’s reputation and balance sheet more closely than ever to the success of generative AI.

For now, the market has voted to fund the bet. The harder question, whether AI’s eventual profits can justify the debt needed to build it, will be answered over years, not in one record-breaking bond sale.