Oracle reported quarterly revenue of $19.3 billion, up 30%, with an operating profit of $6.7 billion. Capital expenditure reached $28.5 billion for the quarter, against $8.5 billion a year earlier. Interest charges tied to financing this build-out rose 55%, to $1.4 billion. The company launched a fresh round of job cuts in the days that followed.
We document the sector's commitments through the lens of the amounts involved. Here is what they look like in published accounts, and what that implies.
The figure that raises the question
Spending $28.5 billion in capital expenditure in a single quarter against revenue of $19.3 billion means the company is investing roughly one and a half times what it takes in.
That is not unusual for a construction-heavy industry: a power utility or a telecoms operator in a deployment phase shows the same profile. It does, however, assume two things.
External financing. The difference comes from debt, which is why interest charges are up 55%. That is the cost of the bet.
Future demand that materialises. An investment of this scale is justified by revenue expected over a decade. If demand arrives, the assets are profitable. If it is slow to arrive, the debt remains.
The simultaneity with the job cuts
This juxtaposition deserves to be explained rather than denounced, because it is coherent.
A company committing considerable sums to infrastructure will mechanically seek to reduce its ongoing costs. This is not a replacement of employees by machines: it is a shift of budget from staff costs to equipment costs.
This is exactly the explanation we put forward in our article on job cuts, noting that the stated reason is not the cause. Here is the mechanism in published accounts: interest is up 55%, and it has to be financed somewhere.
There is a documented concern in the financial sector, and it rests on three points. The circularity of the financing, when the same players invest in each other and each investment inflates the other's revenue. The timeline to profitability, which nobody states precisely. And the possibility that cheap open models compress prices before the investments are amortised. These three concerns do not say that a collapse is coming; they say that the model rests on assumptions the accounts do not yet demonstrate.
What this confirms
These figures concretely validate what we wrote on Friday about the $517 billion in disclosed commitments.
The business model of this sector rests on an assumption of long-term growth, financed by debt, with assets that depreciate over years. It is the classic profile of an infrastructure industry — railways, electricity grids, telecommunications.
These precedents share two characteristics worth recalling. They all ended up profitable. And they all went through a phase where investors lost a great deal of money before that happened, because construction always preceded usage.
What to take away
These accounts are useful because they are audited, which is rare in what we report.
Most of the figures we cite — announced margins, annualised revenues, disclosed commitments — come from companies that publish them without any obligation to do so. A published quarterly result is legally binding, breaks down its line items, and makes it possible to track a trajectory.
That is why we repeat that an initial public offering in this sector would be the most illuminating event of the year. Not for the spectacle, but because it would turn assertions into documents that can be held against you. We would finally know how much what is being built actually costs, and who is paying for it.