NextFin News - SoftBank Group is preparing a roughly 1 trillion yen ($6.3 billion) corporate bond sale to Japanese retail investors, a deal that would become the largest retail bond issuance ever by a Japanese company and deepen the conglomerate's reliance on household savings to fund its artificial-intelligence spending spree. Terms, including the coupon on the seven-year notes, are expected to be finalized in early September, with the offering marking SoftBank's third retail bond sale this year.
The transaction arrives as SoftBank's funding needs accelerate: the company is increasing its stake in OpenAI, closing an acquisition of ABB's industrial robotics business, and racing to build AI infrastructure while interest rates in Japan climb to their highest levels in three decades. The previous record for a Japanese retail bond — 600 billion yen — was set by SoftBank itself in 2025.
The Deal: Size, Structure, and What the Money Is For
The planned 1 trillion yen offering — about $6.3 billion at current exchange rates — is roughly 2.4 times the size of the previous Japanese retail-bond record and would rank among the largest corporate bond deals of any kind in Japan this year. The bonds carry a seven-year maturity, and market participants expect the coupon to be set in the upper 4% range, consistent with SoftBank's recent retail issues, which have priced at the top end of their guidance as investors demand compensation for duration and credit risk in a rising-rate environment. SoftBank declined to comment on the report, which was based on information the company had not released.
The financing logic is visible in the company's own disclosures. In February, SoftBank announced a $30 billion follow-on investment in OpenAI, payable in three $10 billion tranches on April 1, July 1, and October 1, 2026, at a $730 billion pre-money valuation. Upon completion, SoftBank's cumulative investment in OpenAI is expected to total $64.6 billion, representing an ownership interest of approximately 13%. The company said the consideration would be "financed initially through bridge loans and other financing arrangements from major financial institutions, and subsequently replaced over time through the utilization of existing assets and other financing measures."
"AI is transforming the world at an unprecedented pace. OpenAI is a clear leader, with world-class technology and an unparalleled global user base, and we have strong conviction in its continued growth," Masayoshi Son, SoftBank's chairman and chief executive, said in the February announcement.
The retail bond proceeds are expected to be used to repay existing corporate bonds and to invest in AI-related projects, including the OpenAI stake and the ABB robotics acquisition. That mix — refinancing old debt while adding new commitments — is the crux of the story. SoftBank is not raising money because it lacks assets; it is raising money because its assets are locked into illiquid, long-duration technology bets that cannot be sold quickly without accepting steep discounts.
Why Retail Investors: The Funding Ladder in a Rising-Rate Japan
SoftBank's turn to retail investors is neither accidental nor new. It is the predictable output of a funding ladder that has shifted as the Bank of Japan normalizes policy. On June 16, 2026, the central bank raised its benchmark rate to around 1.0%, the highest level since 1995, in a 7-1 vote at the first regular policy meeting held without Governor Kazuo Ueda, who was absent due to illness; board member Toichiro Asada cast the lone dissenting vote. The yield on the benchmark 10-year Japanese government bond climbed above 2.8% in July, its highest level since 1996. For decades, Japanese households parked savings in deposits yielding essentially zero. Now, for the first time in a generation, corporate bonds offering coupons in the high 4% range look genuinely attractive next to government paper.
The data bear this out. Japanese companies raised a record 16.7 trillion yen through corporate bond issuance in the financial year through March 2026, and the retail segment alone hit a record 2.79 trillion yen in yen-denominated issues in 2025. SoftBank has been one of the most active issuers: in April it priced 418 billion yen of retail bonds at an initial five-year coupon of 4.97%, and in June it returned to the retail market with a 35-year subordinated bond guided at 4.8% to 5.6%. The new 1 trillion yen deal would be more than double the size of either.
There is a second reason SoftBank keeps coming back to households. Retail investors are sticky. Unlike institutional lenders, who can demand covenants, collateral, and mark-to-market discipline, retail bondholders typically hold to maturity and do not monitor the issuer's loan-to-value ratio quarter by quarter. For a company whose balance sheet is dominated by the fair value of listed and unlisted technology stakes, that patience is valuable. SoftBank's financial policy caps its loan-to-value ratio — adjusted net interest-bearing debt divided by the equity value of its holdings — below 25% under normal conditions, with an upper threshold of 35% in emergencies, and it says it holds enough cash to cover bond redemptions for at least the next two years.
But the policy is only as strong as the denominator. When the equity value of SoftBank's holdings falls — as it did when the Vision Fund portfolio repriced — the LTV ratio rises mechanically, forcing asset sales or new equity at the worst possible time. Retail bonds do not solve that problem; they only defer it and change who bears the risk.
The Credit Reality: BB+ and a Widening Risk Premium
SoftBank's credit profile is the counterweight to the growth narrative. Standard & Poor's rates SoftBank Group BB+ — speculative grade — a rating the agency affirmed even after the $30 billion OpenAI commitment, noting the company could limit negative financial impact by selling assets. Japan Credit Rating Agency assigns an A rating, but the S&P assessment is the one that prices in global markets.
The market is charging accordingly. SoftBank's credit risk, as measured by credit-default swaps, remains among the highest for Japanese companies, widening about 70 basis points in 2026. That premium flows directly into the coupons the company must offer: the 4.97% on April's notes and the June subordinated deal's guided 4.8% to 5.6% range were the highest SoftBank had paid on yen-denominated debt in more than a decade and a half. The last time SoftBank sold a senior yen bond with a coupon above 4% was 2009, at 4.52%.
The balance sheet shows why investors demand that premium. As of the end of 2025, SoftBank carried about 16.3 trillion yen in stand-alone interest-bearing debt. In the April-June quarter of fiscal 2026, net profit fell 18% despite gains from its Intel investment, according to company results. The earnings engine is increasingly a function of mark-to-market swings on a concentrated portfolio rather than recurring cash flow.
There is an asymmetry here that retail investors should understand. When SoftBank's holdings rise, the gains accrue to SoftBank's shareholders, not to the bondholder, whose return is capped at the coupon. When the holdings fall, the bondholder's claim on the company's assets becomes more valuable — but only if there are enough unencumbered assets left after the LTV machinery forces sales. A 4% coupon is attractive in Japan; it is also the price of underwriting a venture-capital portfolio wrapped in a bond wrapper.
Is This Cyclical or Structural? The Funding Regime Has Changed
The important question is whether SoftBank's retail-bond strategy is a cyclical response to a temporary funding squeeze or a structural shift in how the company — and perhaps Japanese conglomerates more broadly — will be financed. The answer is both, and confusing the two leads to the wrong conclusion.
The cyclical leg is clear: SoftBank needs bridge financing now because three $10 billion OpenAI tranches are landing on fixed dates, the ABB robotics deal requires cash, and the company is refinancing bonds that mature in a higher-rate world. Once those commitments are funded and if OpenAI lists successfully, the immediate pressure eases. In that sense, the 1 trillion yen issuance is a timing decision, not a permanent change in capital structure.
The structural leg runs deeper. Japan's zero-rate era, which allowed companies to roll debt at negligible cost and let households accept zero-yield deposits, is over. The BOJ is on a tightening path, the 10-year government bond yield is back to levels not seen since the 1990s, and households are being nudged — by policy and by inflation running above the central bank's 2% target — into risk assets. That is a regime change, not a cycle. SoftBank is not just exploiting a temporary window; it is building a durable funding channel into Japan's household savings pool, where cash and deposits alone exceed 1,000 trillion yen and still account for more than half of all household financial assets.
The mechanism, then, is a transfer of risk from the institutional periphery to the domestic household core. Institutions price SoftBank's credit daily and can exit. Households, attracted by a coupon that looks high relative to deposits and government bonds but is modest relative to the underlying risk, will likely hold. If SoftBank's AI bets compound, everyone wins. If they do not, the retail bondholder discovers that "4% yield" was the market's honest assessment of default risk, not a free lunch.
The Counter-Thesis: SoftBank Can Manage This
The strongest case against this reading is that SoftBank's balance sheet is far more robust than the junk rating implies, and that critics are looking at the wrong numbers. Portfolio managers at Comgest have argued that SoftBank's assets still comfortably cover its debt obligations and that the company's loan-to-value ratio remains below 25%. The company itself points to its cash position, which it says covers bond redemptions for at least two years, and to a deep bench of liquid listed holdings — including Arm Holdings, which it took public in 2023 — that can be sold or pledged if needed. SoftBank also sold $3.6 billion of dollar- and euro-denominated bonds to institutional investors alongside its April retail deal, showing it retains access to wholesale markets.
There is force in this argument. SoftBank is not funding consumption or covering operating losses; it is funding a concentrated bet on what management believes is the defining technology transition of the era. If artificial general intelligence arrives on the timeline Son has implied, a 13% stake in OpenAI acquired at a $730 billion pre-money valuation could be worth multiples of its cost, and today's coupons will look trivial in retrospect. Son has said the AI boom will require $5 trillion in annual investment and has called suggestions that AI is a bubble "blasphemy." The bond is simply the tollbooth on the way to that outcome.
But the counter-thesis depends on two assumptions that are not yet proven: that OpenAI can monetize at a scale commensurate with its $852 billion post-money valuation, and that an initial public offering will occur on a timetable that relieves SoftBank's funding pressure before the LTV threshold binds. The first is a technology and competition question — corporate spending data from Ramp showed Anthropic overtaking OpenAI in business AI adoption in 2026, at 34.4% of business AI spending versus OpenAI's 32.3%. The second is a market-timing question that SoftBank does not control. A bondholder cannot eat equity upside; a bondholder needs cash flow on a fixed schedule. SoftBank's cash flow is a function of exits it cannot dictate.
What to Watch: The Signals That Will Prove This Right or Wrong
Three signals will determine whether this issuance is remembered as shrewd bridge financing or as the moment SoftBank overreached. First, the pricing: if the coupon lands in the upper 4% range as expected and the deal is heavily oversubscribed, it confirms that Japanese households remain willing to underwrite SoftBank's credit at current spreads. A failed or upsized-coupon deal would signal the opposite. Second, OpenAI's IPO timetable: if the listing occurs before the third tranche closes on October 1, 2026, SoftBank's funding math improves dramatically; a delay pushes more refinancing into 2027 at potentially higher rates. Third, the LTV ratio: if SoftBank's holdings fall enough to push LTV toward the 35% emergency threshold, the company would be forced into asset sales at depressed prices — the classic downside spiral for leveraged investment vehicles.
The specific falsifying signal for the view that this is manageable: if SoftBank's stand-alone interest-bearing debt rises above 20 trillion yen while the fair value of its listed holdings falls more than 20% from current levels, the LTV buffer erodes faster than asset sales can replenish it, and the BB+ rating comes under genuine pressure. Conversely, if the OpenAI IPO prices above a $1 trillion valuation and SoftBank's LTV stays below 25% through 2027, the bear case collapses.
Outlook: Who Benefits, Who Is Exposed
In the short term, the issuance is a liquidity event, not a solvency event. SoftBank buys itself time and optionality, and Japanese retail investors get a yield they cannot find elsewhere. The beneficiaries are SoftBank's existing shareholders, who avoid dilutive equity issuance, and the company's institutional lenders, whose bridge loans get repaid. The exposed are the retail bondholders, who are being asked to accept venture-style risk with bond-style documentation, and the broader Japanese corporate bond market, which could face supply pressure if other large issuers follow SoftBank into the retail pool.
Over the medium term, the outcome hinges on OpenAI's IPO and the performance of the AI infrastructure buildout. A base case has the bond pricing successfully in the upper 4% range, the OpenAI listing occurring in 2027, and SoftBank refinancing again at a lower spread once the equity value of its stake is realized. An upside case has OpenAI listing above a $1 trillion valuation, SoftBank's holdings rerating sharply higher, and the 1 trillion yen deal remembered as cheap, well-timed funding. A downside case has the IPO delayed into 2028, government bond yields climbing past 3%, and SoftBank forced to issue more debt at higher coupons or sell assets into a weak market — at which point the 4% coupon looks like the first installment of a much more expensive funding program.
The long-term structural question is whether Japan's household savings pool becomes a permanent funding source for corporate risk-taking, the way it once funded the postwar industrial buildout through bank deposits. If it does, SoftBank is not an outlier; it is a pioneer. If it does not, this deal is a one-off made possible by a temporary gap between household yield expectations and market pricing.
SoftBank is betting the gap will close in its favor, and that today's retail bondholders will be happy to have locked in 4% when the AI boom they are funding either pays off or does not. The bond is not a vote on whether artificial intelligence will change the world. It is a vote on whether SoftBank's slice of that change will be large enough, and liquid enough, to pay everyone it owes — and the price of that vote just went up.
Explore more exclusive insights at nextfin.ai.
