Apollo expands $9.2 billion loan to SoftBank to fuel AI investments

The rise of private credit

Apollo has established itself as a primary private lender for SoftBank, a move that highlights a shift in how massive technology investments are financed. The relationship began in 2021 when SoftBank needed to assemble a $4.5 billion deal in a very short 45-day window. While major banks like Goldman Sachs and JPMorgan were options, Apollo won the business by moving faster and structuring a complex deal.

The specific financial tool used is a net asset value (NAV) loan. This type of lending is structured against the stated worth of a fund's holdings rather than just the cash on hand. In this case, the loan was built against more than 150 holdings within SoftBank's Vision Fund 2. Apollo acted as the lead arranger, taking a 1% fee and distributing portions of the deal to other partners.

This arrangement allows SoftBank to access capital that might otherwise be harder to secure from traditional sources. Over the last 15 months, Apollo has increased the size of this facility on two occasions, bringing the total to $9.2 billion. The agreement is set to end in 2029 and features a variable interest rate of 7.3%, which is notably higher than the roughly 4% rate on SoftBank's current debt.

Concentration in OpenAI

A major challenge for Apollo is managing the concentration of risk within the loan's collateral. Because OpenAI has seen such a massive surge in value, it now dominates the Vision Fund 2. The AI company currently accounts for about three quarters of the fund's fair value, which is estimated at $1.25 trillion. This makes the loan highly sensitive to the performance of a single company.

To protect itself, Apollo built specific limits into the NAV loan to spread out the risk. For example, OpenAI cannot represent more than one fifth of the collateral in the loan structure. This cap is intended to provide a buffer in case the valuation of the AI company takes a sudden and sharp hit. Despite these rules, the sheer size of OpenAI's success has reshaped the fund's underlying balance.

The loan-to-value ratio has risen since the deal was first created, though it currently stays below 20%. Apollo also stepped in as the only non-bank lender in August when SoftBank struggled to secure a separate $10 billion loan. That specific facility includes protections for Apollo if OpenAI's value falls sharply, including provisions for early repayment if necessary.

Insurance capital and risk

A significant portion of Apollo's exposure to SoftBank sits on the balance sheet of its insurer, Athene. As of late September, Athene held at least $4.8 billion in SoftBank-related debt. This amount represents about 1% of Athene's gross invested assets, making it one of the insurer's largest single-name positions. This high level of exposure is currently pushing against the company's internal concentration limits.

Apollo is now exporting this specific model of using insurance capital to fund large-scale investments. Last year, the firm signed nearly $20 billion in NAV loans for Athene involving private equity managers like Vista Equity and Warburg Pincus. These deals are designed to give borrowers with lower credit ratings access to larger pools of capital that are typically reserved for investment-grade debt.

However, the rapid growth of this model is meeting some resistance as market uncertainty increases. Apollo executives have acknowledged that recent downturns in AI-related stocks and corporate debt have caused the firm to take a more cautious approach. They indicated it is unlikely they will commit more OpenAI-related capital for SoftBank until the market environment becomes clearer.

The future of AI funding

The broader AI investment landscape is facing growing scrutiny as the costs of building out infrastructure continue to climb. SoftBank recently reported a $7 billion drop in value following a decline in the price of several tech investments, and its shares have fallen more than 50% since June. These factors, combined with rising credit risks, are forcing lenders to look more closely at the underlying economics of AI projects.

The central question for investors is whether these massive capital investments will eventually generate enough cash flow to justify the initial costs. While the demand for data centers and computing power is high, the long-term profitability of the companies building them remains to be seen. If technology valuations deteriorate suddenly, the risk for private credit lenders like Apollo increases significantly.

The pace of private credit funding for the AI buildout could slow if lenders pause or if the assets securing these loans underdeliver on cash flow. This could have a ripple effect across the entire AI supply chain. For now, Apollo is balancing the need to fuel AI growth with the necessity of containing downside risk through careful loan structuring and strict collateral limits.