Tech giants have committed nearly $300 billion in guarantees to finance AI infrastructure, utilizing special purpose vehicles to manage debt off their balance sheets.
Leading technology firms have pledged guarantees totaling approximately $300 billion over the last year to finance data centers and infrastructure for artificial intelligence. Much of this debt is recorded on the balance sheets of special purpose vehicles (SPVs), allowing these companies to maintain a healthier balance sheet while funding their AI initiatives, as reported by the Financial Times.
Major players such as Meta, Alphabet, Nvidia, and Broadcom are leveraging these SPVs to secure financing at favorable credit ratings without directly incurring equivalent debts.
These SPVs typically raise debt and own the physical assets, such as data centers, while the tech firms either lease the capacity or provide guarantees. If the assets are sold or leased for less than agreed, the guarantees cover the difference.
This arrangement enables borrowers to access funds at lower interest rates, which are generally only 1 to 1.5 percentage points higher than the borrowing costs of the guarantor companies.
For instance, Alphabet's obligations related to credit derivatives for its data centers jumped from $16.9 billion to $43.8 billion in a six-month span, as noted in its quarterly report. The fair value of these commitments on its balance sheet stands at $815 million, alongside $7.6 billion in financial guarantees. Additionally, Alphabet has committed to providing $24.1 billion in future guarantees for data center and energy infrastructure development.
Nvidia's Role with OpenAI
Nvidia has also made significant commitments, issuing guarantees up to $105 billion for SB Energy's data centers in Ohio, which will be leased by OpenAI. These guarantees may require Nvidia to assume control of the lease if OpenAI defaults on payments, although this risk will decrease as OpenAI fulfills its payment obligations.
Meta and Broadcom's Contributions
Meta has extended residual value guarantees for its Hyperion data center, with a total exposure of about $28 billion. If Meta does not renew its lease, its maximum liability would be the difference between the asset's fair value and the guarantee threshold, which is expected to be unlikely by the end of 2025. These guarantees have facilitated around $27 billion in debt for Hyperion at interest rates below 1.5 percentage points above Meta's bond yields.
Broadcom is utilizing a similar strategy for AI equipment, recently launching the AI XPV platform with initial funding of $35 billion. Under its lease support agreement, Broadcom's maximum potential liability is approximately $29 billion.
Risks of Off-Balance Sheet Financing
Despite the benefits, analysts at Morgan Stanley have highlighted over $3.1 trillion in off-balance sheet obligations and credit support among seven major tech firms and AI chip manufacturers, indicating that these guarantees carry significant risk. Doug Kolandrea from KBRA noted that the off-balance sheet exposure has markedly increased, complicating the credit risk profiles of these companies.
"This significantly complicates the credit risk profiles of these companies," Kolandrea remarked.
The economic risks associated with these financing schemes remain, particularly if the value of the data centers or equipment falls or if tenants fail to meet their obligations. Rating agencies take these off-balance sheet structures into account when evaluating debt levels, although adjustments have generally been minimal so far.
In September, reports also emerged regarding OpenAI's acquisition of thousands of Mac mini and Mac Studio computers for specific operational tasks.
