The artificial intelligence boom is creating an extraordinary infrastructure spending cycle.
But investors are beginning to ask a difficult question:
How much debt can the AI infrastructure industry support?
That question became harder to ignore this week after around $18 billion of loans tied to an Oracle-leased data center project in New Mexico came under pressure.
The loans were recently quoted at roughly 89 to 91 cents on the dollar, according to Reuters.
That does not mean Oracle has defaulted.
It does, however, show that investors are beginning to scrutinize the economics and financing structures behind the enormous infrastructure investment required to power AI.
Why It Matters
AI companies need computing power.
Computing power requires data centers.
Data centers require land, electricity, cooling systems, networking equipment and billions of dollars in capital.
And that money increasingly comes from debt.
Oracle has become one of the most aggressive companies in this infrastructure race.
The company has been expanding its cloud infrastructure to serve major AI customers, including OpenAI.
But the scale is becoming enormous.
Oracle has forecast capital expenditure of up to approximately $95 billion for fiscal 2027.
That number alone explains why investors are starting to look beyond AI revenue growth and ask another question:
What is the return on all this infrastructure spending?
The Details
The $18 billion financing is connected to a large data center project in New Mexico that Oracle has leased.
Loans associated with the project trading below face value is significant because debt markets are effectively assigning a higher risk premium to the financing.
A loan trading at 90 cents on the dollar does not automatically mean the underlying project is failing.
Debt prices can move for many reasons.
But it does indicate that lenders and investors are paying closer attention to the financial structure behind AI infrastructure.
The AI Infrastructure Equation
The economics of AI infrastructure are relatively simple.
Companies spend billions today to build capacity.
They then need customers to pay enough over many years to generate an acceptable return.
The problem is that AI technology itself is changing extremely quickly.
A data center designed around today’s demand assumptions has to remain economically useful for years.
That creates an unusual risk.
The technology is moving at software speed.
The infrastructure is built at industrial speed.
That mismatch matters.
AI models can improve every few months.
A multi-billion-dollar data center cannot be rebuilt every few months.
Oracle’s Spending Is the Bigger Story
The $18 billion financing issue is interesting.
Oracle’s overall spending plans are even more important.
Capital expenditure approaching $95 billion represents an enormous financial commitment.
The company needs AI demand to continue expanding at a very high rate to justify that level of investment.
This is where the AI boom becomes a business story rather than simply a technology story.
If AI demand keeps growing rapidly, the infrastructure could become extremely valuable.
If demand growth slows, companies could find themselves with huge fixed costs.
Is This an AI Bubble?
It is too early to say that the entire AI infrastructure market is a bubble.
There is genuine demand.
Companies are using AI.
Cloud providers are expanding capacity.
AI models are becoming more capable.
But there is a difference between real demand and economic returns on every dollar of infrastructure investment.
That distinction is becoming increasingly important.
The internet boom provides a useful historical comparison.
The internet was real.
The technology transformed the economy.
But that did not mean every company that invested heavily in internet infrastructure generated attractive returns.
AI could follow a similar pattern.
The technology can be transformative while some investments still turn out to be uneconomic.
What This Means for the Industry
The next stage of the AI boom may therefore be less about who can spend the most and more about who can spend the most efficiently.
That changes the competitive landscape.
Companies need to answer:
- How much does each AI workload cost?
- How much revenue does each data center generate?
- How quickly can capacity be filled?
- How long will the equipment remain competitive?
- How much debt is required?
- What happens if AI demand grows more slowly?
Those questions matter just as much as benchmark scores.
The Hidden Risk: Financing
The most important development may actually be happening outside the AI labs.
Banks, bond investors and infrastructure financiers are increasingly becoming part of the AI ecosystem.
As long as AI demand keeps rising, financing can support more data centers.
But if the expected returns deteriorate, lenders become more cautious.
That could slow the expansion of AI infrastructure even if demand for AI itself remains strong.
What’s Next?
The market will be watching Oracle closely.
Investors will want to see whether the company’s enormous AI infrastructure spending translates into sustained revenue and cash flow growth.
The broader AI industry will face the same test.
The first phase of the AI boom was about proving that AI works.
The second phase is about proving that AI infrastructure can make money.
That may be a much harder test.
FAQ
What is the $18 billion Oracle data center issue?
Around $18 billion of loans associated with an Oracle-leased data center project in New Mexico have come under pressure, with the loans quoted below face value.
Has Oracle defaulted on the debt?
No. Loans trading below face value does not mean a default has occurred.
How much is Oracle planning to spend on infrastructure?
Oracle has forecast capital expenditure of up to approximately $95 billion for fiscal 2027.
Why does AI require so many data centers?
AI models require enormous computing power. Training and running these systems requires specialized chips, networking, cooling and electricity.
Does this mean the AI boom is a bubble?
Not necessarily. AI demand is real, but the profitability of individual infrastructure investments remains an important question.
Related reading:
- Why AI Data Centers Are Becoming the New Oil Fields
- The $1 Trillion AI Infrastructure Race
- Why Nvidia Is So Important to the AI Economy
