The race to build artificial intelligence infrastructure is leaving behind a figure that many investors overlook. An analysis by Nikkei Asia, based on the quarterly reports of Alphabet, Amazon, Meta, Microsoft, and Oracle, estimates that these companies have accumulated around $1.65 trillion in off-balance sheet commitments, exceeding the $1.35 trillion of debt officially reflected on their balance sheets. This is not hidden debt in the accounting sense nor an irregular practice but future obligations primarily arising from contracts for data centers, servers, and AI computing capacity that have not yet gone into effect.
The key points of AI-related off-balance sheet commitments in 20 seconds
- The five major US tech giants together hold about $1.65 trillion in future commitments related to AI.
- Most of this stems from long-term contracts for data centers, GPUs, servers, and processing capacity.
- These commitments appear in the financial notes, not as debt on the balance sheet, because many contracts are not yet active.
- If AI demand does not grow as expected, these companies will still need to cover these costs once the facilities come online.
Although some headlines refer to “hidden debt,” that phrase can be misleading. In reality, these obligations are disclosed in financial statements and comply with current accounting standards. The difference is that they do not yet qualify as liabilities on the balance sheet because many data centers are not operational or the contracted hardware has not yet been delivered.
The bill for securing capacity for AI
The expansion of artificial intelligence has changed how hyperscalers plan their investments. Instead of waiting for capacity to become available, companies like Microsoft, Amazon, Google, and Meta are signing multibillion-dollar, multi-year contracts to secure electricity, buildings, GPUs, servers, and processing capacity even before new data centers are operational.
This model reduces the risk of infrastructure shortages in a market where demand for computing still grows but also involves taking on enormous financial commitments for many years.
According to Nikkei’s study, Meta would hold around $420 billion in such obligations, while Oracle has increased its commitments to about $273.3 billion, driven by large-scale AI and data center projects.
This situation does not imply liquidity problems for these companies or non-compliance with accounting rules. It simply reflects that a significant portion of their committed investments remains off the balance sheet because the execution of many contracts will begin in the future.
A risk that depends on future demand
The real debate is not about the legality of the accounting treatment but about the economic risk these commitments pose.
When new data centers come online, companies will need to start paying for the contracted capacity regardless of whether there is sufficient demand to utilize it. If AI growth maintains its current pace, these investments could be recouped through new cloud services, AI models, and enterprise clients.
In fact, Alphabet, Amazon, and Microsoft have a combined cloud service portfolio valued at around $1.45 trillion, which supports their investments. Amazon Web Services has also publicly defended that its investments are not speculative bets.
However, the scenario changes if market growth slows down.
In that case, hyperscalers might face overbuilt infrastructure and long-term contracts that continue generating costs even if data center utilization falls below expectations.
Additionally, some analysts are beginning to monitor another element: several companies are financing a large part of these investments through corporate debt issuance and new equity, while infrastructure spending grows faster than their profits.
Not a new Enron, but a signal worth understanding
Some analyses have inevitably recalled Enron’s history because of the term “off balance sheet.” However, caution is necessary in making comparisons.
In Enron’s case, off-balance sheet structures were used to hide losses and problematic assets through shell companies, which ultimately led to fraud.
The current situation is different. The commitments of Alphabet, Amazon, Meta, Microsoft, and Oracle are documented in financial disclosures, are known by auditors, and follow accepted international and US accounting standards. The concern for part of the market is not their existence but that many investors focus solely on the debt reflected on the balance sheet without considering the magnitude of future obligations already committed.
In upcoming quarters, it will be especially important to watch two indicators: the actual demand for AI services and how quickly the billions invested in new data centers are brought into operation. Both will determine whether these commitments become a competitive advantage or a financial burden.
Frequently Asked Questions
What does it mean for an obligation to be off-balance sheet?
They are future commitments that do not yet meet the accounting criteria to be recorded as debt on the balance sheet, although they are disclosed in the notes to the financial statements.
Are Big Tech companies hiding debt?
Not in terms of accounting fraud. These commitments are disclosed according to applicable regulations, though their off-balance sheet presentation may make their volume less obvious to some investors.
Why are these obligations related to artificial intelligence?
Primarily because they stem from long-term contracts to build and operate data centers, acquire servers, GPUs, and secure the necessary computing capacity for AI services.
What is the main risk?
If AI service demand grows less than expected, companies will still be obliged to pay for much of the infrastructure once contracts are active, regardless of utilization.
via: tomshardware
