Two days before SoftBank Group presents its Q2 FY2026 results on November 10, the bond market is already offering a verdict. The company’s five-year credit default swap spread climbed to 466 basis points last Friday, its highest level since October 2022. At that level, Reuters reported that market data implied a 30.6% probability of default. The question investors must settle before the briefing is whether that number reflects a real financing crisis or an overreaction to complexity.
The Bull Case: Arm Is Real, and the Mark Will Hold
The bear argument on SoftBank leans heavily on the OpenAI stake being an illiquid, paper gain on a private company. That framing misses what has actually happened to the underlying asset. SoftBank completed the final $10 billion tranche of its follow-on OpenAI investment on October 1, lifting cumulative investment to $64.6 billion for a reported 13% ownership interest.
The most recent disclosed valuation figure investors can anchor to is still SoftBank’s own: in its FY2025 results materials, the company put the fair value of its OpenAI stake at $79.6 billion as of March 31, 2026, against a cumulative cost of $34.6 billion.
Arm Holdings, SoftBank’s listed anchor, is delivering the kind of results that give the bull case structural backbone. Arm delivered its fifth consecutive quarter of revenue growth above 20%, with total revenue rising 27% year-over-year to $1.44 billion, its best-ever second quarter. Arm’s first in-house data center chip, the AGI CPU, now carries more than $2 billion of expected revenue across fiscal years 2027 and 2028, and Arm has reiterated its longer-term forecast of $15 billion for that business. That is a tangible, listed asset SoftBank can borrow against, and has.
On the bridge loan itself, the maturity is tighter than it looks for a company with options. SoftBank’s March 2026 announcement of the facility states the group expects to repay the bridge facility in stages by its March 25, 2027 maturity through existing assets and other financing measures. If OpenAI prices an IPO in the late-2026 window at a valuation that supports SoftBank’s marked value, the stake converts to tradable equity and the bridge facility is refinanced or repaid into a cleaner capital structure.
The Bear Case: The Clock Is Real, and the Mark May Not Be
The CDS market is not simply panicking. It is pricing a specific sequence of events that must occur for SoftBank to avoid refinancing distress, and that sequence has already started slipping. SoftBank arranged an unsecured $40 billion bridge facility in March 2026 to support its OpenAI follow-on investments, with a March 25, 2027 maturity.
That pressure on SoftBank’s balance sheet sits alongside a parallel capital raise. Masayoshi Son has been pitching Gulf investors on a new fund to acquire and retool companies using AI, which signals how much fresh external capital the group believes it needs.
But the near-term refinancing cliff in this debate is not as clean as it sounds, because SoftBank has already moved to shrink it. In its October 1, 2026 update on the OpenAI follow-on, SoftBank said it had repaid all borrowings under the bridge facility and canceled the remaining undrawn commitment.
That does not eliminate financing risk. Virtually none of SoftBank’s OpenAI gains are cash. They are paper profits on a private company whose valuation is still ultimately set by private rounds and management marks. Reuters has reported that OpenAI has delayed its public offering to 2027, which keeps SoftBank’s liquidity story tied to debt markets and asset-backed borrowing rather than an imminent public exit. If public markets balk at a $1 trillion valuation for OpenAI, SoftBank will either have to refinance at tougher terms or the company will have to accept a lower market cap. A lower mark directly reduces the investment gains line that has been a meaningful contributor to SoftBank’s reported results.
SoftBank’s market value halved from its June peak as higher borrowing costs fuel investor concerns over the sustainability of its AI investments. Reuters also reported that the company raised $11.1 billion in September via the largest corporate junk bond sale on record to finance Masayoshi Son’s AI ambitions, which signals the refinancing door is open but not cheap.
Where the Evidence Leads
The bear case is better supported right now, but not for the reason most commentators give. The risk is not that Arm or OpenAI are worthless. The risk is timing and volatility: how quickly private marks and public prices can swing, and how those swings interact with SoftBank’s funding costs. For SoftBank, the real test may not be whether AI ultimately creates wealth, but whether the pace of asset-price swings can match the schedule for paying debt interest.
Monday’s earnings briefing will force SoftBank to put a fresh mark on the OpenAI stake under conditions materially worse than the last report. If that mark holds above cost, the bull case gains credibility. If it shows compression toward the $64.6 billion invested, the refinancing math becomes the story. Watch the investment gains line and any language around near-term funding and takeout financing. Those two items will settle this debate more than any CDS spread can.
