NEW YORK: The artificial intelligence revolution is entering a new financial phase.

For years, technology companies largely relied on enormous cash reserves and strong operating profits to finance their AI expansion. Now, the cost of building data centres, buying advanced processors and securing computing capacity has become so large that companies are increasingly turning to borrowed money.

That shift is creating a new question for investors:

Could the AI boom become a debt boom — and what happens if the expected profits arrive later than the loans come due?

Why Does AI Need So Much Money?

Artificial intelligence may look like a software revolution, but modern AI requires enormous physical infrastructure.

Companies need:

• Advanced AI chips
• Data centres
• Electricity and cooling systems
• High-speed networking equipment
• Storage infrastructure
• Cloud computing capacity
• Engineers and specialised technical systems

The most advanced AI models require huge amounts of computing power during both training and everyday use.

As competition has intensified, technology companies are racing to secure computing capacity before their rivals do.

That has transformed AI from primarily a software investment into one of the world’s largest infrastructure spending programmes.

Why Are Companies Borrowing Instead of Using Cash?

The simple answer is scale.

The amount of money required for AI infrastructure is becoming too large to rely exclusively on existing cash flows.

Major technology companies are already spending hundreds of billions of dollars on AI-related infrastructure. Combined spending by the largest technology companies is expected to exceed $730 billion this year, according to recent industry estimates.

Borrowing allows companies to build infrastructure immediately rather than waiting years to accumulate enough cash.

The strategy is based on an important assumption:

AI-generated revenue in the future will be large enough to justify today’s investment.

What Is Happening in the Market Right Now?

The latest developments show just how quickly the financing model is changing.

Broadcom is reportedly seeking more than $50 billion in financing connected to AI chip development.

SpaceX has discussed a financing package of around $40 billion, including investment-grade debt and loans, to acquire Nvidia chips.

Oracle is also discussing financing arrangements for major AI-chip purchases.

These deals illustrate how AI infrastructure is increasingly being funded through complex combinations of corporate bonds, bank loans and private financing.

At the same time, Alphabet’s autonomous-driving company Waymo has secured a $5 billion term loan to accelerate its expansion, showing that debt financing is spreading across technology businesses beyond traditional AI laboratories.

What Is an AI Debt Boom?

An AI debt boom occurs when companies borrow increasingly large amounts of money to finance AI-related assets and infrastructure.

Instead of paying the entire cost from cash, a company can borrow money today and repay lenders over several years.

This can be perfectly reasonable when the investment produces predictable future cash flow.

But AI infrastructure presents a different challenge.

Technology changes extremely quickly.

A chip purchased today may become less competitive before a long-term loan has been fully repaid.

That creates a potential mismatch between the lifetime of the equipment and the lifetime of the debt.

Why Are Investors Becoming Nervous?

Debt creates an obligation that does not disappear if business conditions deteriorate.

If an AI project performs exceptionally well, shareholders can benefit from rising profits.

But if an expensive AI project generates less revenue than expected, the company still has to make its interest and principal payments.

That creates an asymmetric risk for lenders.

Recent market analysis shows that investors are demanding greater compensation from riskier AI-related borrowers. AI-related issuance in lower-rated U.S. credit markets has grown sharply, while borrowing costs for weaker companies have increased.

What Happens If AI Profits Do Not Grow Fast Enough?

This is the biggest question facing investors.

Imagine a company spends $20 billion building computing infrastructure.

It expects AI customers to generate enough revenue to cover operating costs, interest payments and eventually repay the investment.

But suppose demand grows more slowly than expected.

The company could then face:

Lower revenue + high operating costs + large debt payments.

That can squeeze profits and potentially force companies to reduce spending, sell assets or raise additional capital.

In extreme cases, heavily indebted companies could face serious financial stress.

Why Are Interest Rates Important?

The timing of the AI debt boom is particularly important because global borrowing costs are already under pressure.

U.S. Treasury yields have moved close to their highest levels in roughly 24 years, while investors are demanding greater returns for holding long-term debt.

When interest rates rise, borrowing becomes more expensive.

A company that could previously finance a project at a relatively low rate may now have to pay substantially more.

For businesses borrowing tens of billions of dollars, even a small change in interest rates can translate into billions of dollars of additional financing costs over time.

Could AI Debt Affect Government Borrowing?

Potentially, yes.

Governments are also heavy borrowers.

When large technology companies simultaneously enter bond markets to raise enormous amounts of capital, they compete with governments and other corporations for investors’ money.

If investors demand higher returns, borrowing costs can rise more broadly.

This is particularly important at a time when governments in several major economies are already dealing with large budget deficits.

The result could be a competition for capital between AI infrastructure, governments and other businesses.

Is This Another Dot-Com Bubble?

The comparison is tempting, but the situations are not identical.

During the dot-com era, many companies had extremely weak or nonexistent business models.

Today’s AI leaders include companies with enormous revenues, established customers and significant cash generation.

The AI technology itself is also already producing real commercial applications.

However, the current concern is different.

The question is not necessarily whether AI is useful.

The question is:

How much can companies spend on AI infrastructure before the financial returns justify the investment?

Why Are Data Centres at the Centre of the Story?

Data centres are the physical foundation of the AI economy.

They require huge amounts of:

• Electricity
• Land
• Cooling
• Servers
• Chips
• Networking equipment
• Construction capital

Building these facilities can take years and require enormous upfront investment.

Because of that, financial institutions have increasingly become important participants in the AI expansion.

Private-credit firms, banks and bond investors can provide capital long before the infrastructure begins generating its full economic return.

What Could Go Right?

The debt strategy could ultimately prove highly successful.

If AI adoption accelerates, companies could generate enormous new revenues from:

• AI software
• Cloud computing
• Enterprise automation
• Autonomous vehicles
• Robotics
• AI agents
• Data services
• Advanced semiconductor products

In that scenario, today’s borrowing could help companies establish dominant positions before competitors.

Debt would then function as a tool for accelerating growth.

What Could Go Wrong?

The opposite scenario would be more complicated.

AI demand could grow more slowly than expected.

Chip prices could decline rapidly.

New technologies could make existing hardware obsolete.

Data-centre construction could become more expensive.

Electricity shortages could limit expansion.

Interest rates could remain high.

Or companies could simply build more computing capacity than customers actually need.

Any combination of these factors could reduce returns on AI investments while leaving companies with large debt obligations.

What Should Investors Watch?

Several indicators will become increasingly important.

First, AI revenue growth. Companies must eventually demonstrate that enormous infrastructure spending is producing comparable revenue.

Second, corporate debt levels. Investors will want to know how much borrowing is being used to finance AI expansion.

Third, interest costs. Rising rates could make heavily leveraged projects less attractive.

Fourth, data-centre utilisation. Empty or underused facilities would make debt-financed expansion particularly risky.

Fifth, chip economics. Rapid technological improvements could reduce the value of older AI hardware.

The Bigger Picture

The AI revolution is no longer simply a technology story.

It has become a capital-markets story.

The next stage of artificial intelligence will depend not only on better models and faster chips, but also on whether companies can finance the enormous infrastructure required to operate them.

Borrowing can accelerate technological progress, but it also increases financial risk.

That is why today’s AI debt boom deserves close attention.

The critical question for the global economy is no longer simply “How powerful will AI become?”

It is also:

“Will the economic returns from AI arrive quickly enough to justify the billions being borrowed today?”