SoftBank Borrowed $40 Billion to Bet on OpenAI. The 12-Month Term Is the Real Signal.

SoftBank announced on March 27, 2026 that it has secured a $40 billion unsecured bridge loan to fund its $30 billion follow-on investment in OpenAI. The loan syndicate includes JPMorgan Chase, Goldman Sachs, Mizuho Bank, Sumitomo Mitsui Banking Corp, and MUFG Bank. The facility matures in March 2027.

The $40 billion figure is the headline. The structure of the loan is what invites closer reading. Five major global banks agreed to provide an unsecured, 12-month facility to SoftBank specifically to fund an investment in a company that has not yet gone public. That structure is unusual, and a short maturity on a loan this size can imply several different liquidity paths, not only an IPO: it could reflect an expectation of refinancing into longer-term debt, a planned asset sale elsewhere in SoftBank’s portfolio, or, yes, an anticipated OpenAI IPO. This analysis treats an OpenAI IPO within the loan term as the most discussed hypothesis in market commentary, not as something the loan documents themselves establish.

How Bridge Loan Economics Work

A bridge loan is a short-term debt instrument typically intended to be repaid from proceeds of a future event, which can include an IPO but can also include refinancing, asset sales, or other liquidity events. SoftBank’s $40 billion bridge loan has a 12-month maturity, meaning the full $40 billion must be repaid or refinanced by March 2027. The loan is unsecured, which means the lenders have no collateral backing the debt. If SoftBank cannot repay, the lenders’ recovery depends on SoftBank’s general corporate assets and cash flow.

The bridge loan’s interest rate has not been publicly disclosed, but unsecured corporate debt of this size and duration typically carries SOFR plus 150 to 300 basis points. At current SOFR rates (~4.3%), SoftBank would be paying approximately 5.8% to 7.3% annually on $40 billion at that typical range, which would translate to $2.3 billion to $2.9 billion in annual interest expense; this is a market-rate estimate, not SoftBank’s disclosed figure.

The Loan Structure and What It Might Imply

SoftBank’s Balance Sheet Under Pressure

SoftBank CFO Yoshimitsu Goto acknowledged the company’s Loan-to-Value ratio could “temporarily” breach its self-imposed 25% ceiling. That ceiling has been SoftBank’s key discipline metric since Vision Fund losses. Breaching it signals the OpenAI bet is consuming borrowing capacity management considers outside normal parameters.

SoftBank’s total investment in OpenAI approaches $60 billion, making it the single largest bet in SoftBank’s history. The Vision Fund’s previous largest bet, WeWork, resulted in a $11.5 billion write-down. Masayoshi Son is concentrating the company’s balance sheet on a single AI company to a degree that creates existential exposure.

The OpenAI IPO Math (One Scenario Among Several)

The Lender Syndicate and What It Might Reveal

The five banks in the syndicate (JPMorgan Chase, Goldman Sachs, Mizuho, SMBC, and MUFG) each took roughly equal shares of the $40 billion facility. JPMorgan and Goldman are the two most active investment banks in technology IPOs, which makes them plausible, though unconfirmed, candidates to underwrite an eventual OpenAI IPO, a role that would generate substantial underwriting fees if it happens. (For comparison, Harvey reached $11 billion in AI’s application layer on far less capital.) One reading of the bridge loan is that it functions partly as a relationship-building step toward an eventual IPO mandate, though neither bank has confirmed such a connection publicly, and banks routinely extend credit to major clients without a guaranteed follow-on mandate.

Mizuho, SMBC, and MUFG are SoftBank’s traditional Japanese banking partners. Their participation reflects the relationship banking model that governs Japanese corporate finance: SoftBank’s main banks support its strategic bets in exchange for the full banking relationship. The $40 billion loan is partly a strategic investment decision and partly a relationship maintenance cost. (The full cost structure of running AI at scale in 2026 adds context to why these bets are so large.)

What Happens If an IPO Does Not Happen

One scenario worth naming explicitly, since it is the flip side of the IPO hypothesis: what if OpenAI does not IPO within the loan’s term, whether or not that was ever the intended repayment path? SoftBank would need to refinance the bridge loan, likely at higher rates and potentially with collateral requirements. One plausible collateral candidate would be SoftBank’s OpenAI shares themselves, which would create a reflexive risk: if a delay signals that OpenAI’s valuation is softening, the collateral value would decline at the same time the refinancing need increases.

SoftBank could also sell other Vision Fund assets to repay the bridge loan, consistent with its own stated plan to repay “partly through the sale of assets,” but the Vision Fund’s portfolio has already been significantly marked down from its peak. The Arm Holdings stake (worth approximately $150 billion) could theoretically cover the bridge loan, but selling Arm shares to fund an OpenAI bet would concentrate SoftBank’s portfolio even further. Several of the available options in a delayed-liquidity scenario carry real costs.

How likely this scenario is depends on OpenAI’s own IPO timeline, which is not something SoftBank’s loan terms control or confirm. This asymmetry, meaningful downside if liquidity is delayed, is part of what makes the loan worth scrutinizing regardless of which specific repayment path SoftBank intends. Masayoshi Son has historically been comfortable with asymmetric bets of this kind. The WeWork experience suggests the downside scenario, while not the base case, is not impossible.

What This Means for the AI Industry

The bridge loan creates a plausible incentive for SoftBank to want faster liquidity from its OpenAI stake, whether via IPO or another path, that did not previously exist in as concrete a form. If SoftBank’s leadership does favor an IPO as the resolution, its stake gives it real influence, but that influence operates alongside Microsoft, Thrive Capital, a16z, and Sequoia, whose own IPO preferences are separately determined and not established by this loan.

For the AI industry as a whole, SoftBank’s bridge loan is worth watching as a signal regardless of exactly how it resolves. The largest non-operator investor in AI just staked $40 billion on a 12-month financing structure tied to an investment in the industry’s leading company (which doubled its workforce to 8,000). If OpenAI does IPO within that window at a strong valuation, it would validate the current valuation framework and likely accelerate capital flows into the sector. If liquidity is delayed or the valuation softens, it could tighten lending to AI companies broadly. Which of these plays out is genuinely unresolved as of this writing.

For Masayoshi Son, the $40 billion loan continues a recognizable Vision Fund pattern: concentrated, debt-amplified bets before what Son is wagering will be a peak valuation. The Alibaba win and WeWork loss were both products of a similar approach. Son appears to be betting OpenAI is the Alibaba moment for AI, an inference from the scale and structure of the bet rather than a stated company thesis. The 12-month bridge is the financial structure that gives him runway to find out, on whatever timeline the resolution actually takes.

Disclaimer: Market context for founders and builders, not financial advice. Sources: TechCrunch, Bloomberg, Reuters. Updated 2026-08-18: reframed the OpenAI-IPO connection throughout as this publication’s leading hypothesis for the loan’s short maturity, rather than an established fact, since the loan documents themselves do not specify an IPO as the repayment mechanism and SoftBank has publicly cited asset sales as part of its repayment plan.