SoftBank’s proposed OpenAI bond financing puts borrowing costs in focus

SoftBank Group’s next OpenAI payment is bringing its investment strategy into the bond market, where lenders will put a price on the time between committing capital and receiving returns.

Reuters reported September 21 that SoftBank had launched US$10 billion and €1 billion of senior unsecured notes, citing a term sheet. Pricing was scheduled for September 24 and settlement for September 29.^[1] This report uses information reviewed through September 22; the proposal is not presented as a completed sale.

An investment deadline comes before the investment return

SoftBank’s July 1 disclosure confirms that it completed a second US$10 billion installment of a US$30 billion additional investment in OpenAI Group PBC through SoftBank Vision Fund 2. It scheduled the third US$10 billion installment for October 1, Japan time.^[2]

Japan.co.jp’s analysis is that the financing question concerns timing as much as conviction about AI. The investment may take years to produce realizable proceeds. Bond interest and principal, by contrast, fall due under contractual schedules. An increase in an investment’s reported value is not itself cash available to service debt.

The borrowing cost remains a question for final pricing

There is a relevant, but separate, precedent. In April, SoftBank announced dollar senior bonds with coupons of 7.625%, 8.250% and 8.500% for 3.5-, 5.5- and 10-year terms, respectively.^[3] Those historical terms are neither guidance nor a forecast for the September transaction.

A coupon determines interest on face value; an investor’s yield also depends on the purchase price and repayment terms. The issuer’s cost can additionally include fees and currency hedging. A lower euro coupon does not automatically make euro borrowing cheaper once the funding is compared in a common currency.

A credit facility is not the same as outstanding debt

SoftBank announced a US$40 billion bridge facility on March 27, with a final repayment date of March 25, 2027.^[4] A facility limit measures available borrowing capacity, not necessarily the amount drawn.

On September 9, the company said it had decided to repay the US$25.9 billion outstanding under that facility on September 15 and had notified the lenders.^[5] That announcement describes a repayment decision. It does not establish the group’s complete debt position on September 22.

These distinctions matter when assessing exposure. Replacing an existing loan with a bond changes funding sources and potentially maturity risk without necessarily increasing debt by the bond’s full face value. Funding a new investment is different. Adding facility ceilings, historical balances and new bond proposals together would overstate what is known.

The useful financial picture reconciles actual drawings, repayments, issuance proceeds and cash deployment. It also distinguishes parent-level obligations from consolidated debt and cash that may sit in subsidiaries. This article does not infer a current total from the transaction announcements.

Size alone cannot establish investor appetite

A large proposed offering tells readers how much the issuer wants to raise. Evidence of demand comes from the final price, the spread above an appropriate same-currency benchmark and the persistence of orders as terms change.

An announced order book, if one becomes available, should not automatically be equated with completed purchases. Similarly, a high coupon can reflect underlying market rates as well as issuer credit risk. It does not by itself prove that a sale struggled.

For unsecured creditors, protections also depend on the contractual terms and the issuer’s ability to generate or mobilize cash. Financing an OpenAI stake does not give a SoftBank bondholder direct ownership of that stake or the same upside as an equity investor.

What the next disclosures need to resolve

The decisive details are the final coupons and issue prices, amounts due at each maturity, the relationship between proceeds and any loan repayments, and liquidity remaining after the investment payment. Those figures will allow a more useful assessment than the headline fundraising total.

SoftBank’s ability to obtain funding and its ability to earn returns above the resulting cost are related but separate tests. A successful bond sale would address the first; investment performance and future cash availability will determine the second.