The bomb hidden in the footnotes of AI giants
Special update
In the next two weeks, Google, Microsoft, Amazon, and Meta will publish their financial reports.
Most investors will look at revenue, earnings per share, and forward guidance...
But there is one number that will likely not appear in the headlines,
even though it could be even more significant than all the financial results of the giants!
An investigation by the media giant Nikkei found that five technology giants – Google, Microsoft, Amazon, Meta, and Oracle – hold off-balance sheet commitments totaling approximately $1.65 trillion.
For comparison, the debt they officially report stands at only about $1.35 trillion.
Before you panic: this is not fraud and not something illegal.
The reason is simple,
When a company signs a contract to lease a data center for ten or fifteen years, or commits to purchasing huge quantities of AI chips from Nvidia and other partners, it is not yet required to record the entire commitment as debt on the balance sheet.
Accounting rules allow it to recognize a large part of the commitments only when the data center begins operation or when the asset comes into use.
In other words, the company has already committed to spending the money, but from an accounting perspective, it does not yet appear as 'regular' debt.
This is why there is a huge gap between the debt seen on the balance sheet and the actual commitments the companies have already signed.
The numbers themselves are enormous:
Meta has off-balance sheet commitments of approximately $420 billion, almost three times the debt it presents on its balance sheet.
For Oracle, the story is more extreme: the volume of commitments surged 30-fold within four years, following the aggressive race to build AI infrastructure.
So why is this important for investors?
Because anyone who only looks at the 'total debt' line on the balance sheet might think the companies are leveraged far less than they actually are.
These commitments exist,
they are simply found mostly in the footnotes of the reports and not on the first page.
It's a bit like a person who takes out a large mortgage, signs a lease contract for a car, and commits to several more future loans.
Currently, they are not yet deducted from the account each month, so their situation looks excellent,
but once all the commitments begin to take effect, the picture changes completely.
Of course, none of this necessarily means there is a problem here.
If the AI revolution continues to grow at the pace companies expect, new data centers will fill with customers, revenue will increase, and these enormous investments could pay off handsomely.
But if demand is lower than forecasts, the risk begins to rise, and then problems will surface.
Data centers that cost hundreds of billions of dollars may not generate the expected return, companies may recognize impairments, and commitments currently off-balance sheet will become real expenses on the books.
It is important to understand that those who financed the construction are also exposed to this,
A large part of the projects was financed through private credit, financing contracts, and project bonds.
Therefore, if something goes wrong, the impact will not stop only with the technology giants, it could also reach investors and financial institutions that funded the race to build AI infrastructure.
This means that the upcoming reports will tell us how much money the companies earned in the last quarter, but to understand the full picture, one must also look at the commitments that have already been signed and have not yet found their way onto the balance sheet.
In the upcoming reports of the giants, the important story will not be found in the headline, but rather in the footnotes.
Keep your eyes open and be alert.