To The Moon

The AI investment boom is entering a new phase, and the latest developments in the US bond market suggest that the cost of building the next generation of AI infrastructure is becoming increasingly difficult to ignore.

On Monday, Oracle led 17 companies issuing bonds in the US high-grade corporate bond market. This was the highest number of issuers entering the market in a single day since September 8. Dealers also expect total bond issuance this week to reach approximately $45 billion.

While companies often accelerate financing ahead of major economic events to avoid market uncertainty, the timing and scale of this borrowing raise a bigger question.

How much debt will technology companies need to fund the AI revolution, and will the returns eventually justify the investment?

For the past two years, investors have focused heavily on the potential of AI. Semiconductor manufacturers, cloud providers and data-centre operators have benefited from enormous capital expenditure plans.

However, building AI infrastructure requires substantial upfront investment. Companies need to secure land, electricity, specialised chips, servers, cooling systems and networking equipment before they can generate meaningful returns.

The spending is happening now, but the financial benefits may take years to fully materialise.

Here are five developments I think investors should watch.

1. Oracle Is Becoming an Important Test Case

Oracle is particularly interesting because of its growing role in AI cloud infrastructure.

The company has been expanding its data-centre capacity to meet demand from major AI customers. Its cloud infrastructure business offers significant growth potential, but scaling this business requires enormous capital investment.

The challenge is that infrastructure spending does not immediately translate into cash flow.

Companies must finance construction, purchase equipment and secure electricity capacity before the infrastructure can generate revenue.

If customer demand continues growing, these investments could eventually produce substantial returns. But if AI demand slows or customers delay spending, companies could find themselves carrying expensive infrastructure and substantial debt obligations.

Oracle’s financing activity therefore deserves attention, not simply because it is issuing bonds, but because it illustrates the capital requirements behind the AI expansion.

2. AI Infrastructure Is Becoming a Debt-Financed Growth Story

Until recently, much of the AI investment narrative centred on the enormous cash reserves of technology giants.

Companies such as Microsoft, Alphabet, Amazon and Meta have considerable financial resources to fund infrastructure expansion.

However, as AI spending increases, even large companies may look to debt markets to preserve financial flexibility and diversify their funding sources.

Borrowing is not necessarily a warning sign. Companies regularly issue bonds to finance long-term investments, particularly when they can access relatively attractive financing rates.

The concern arises when borrowing grows faster than the cash-generating capacity of the underlying business.

If AI infrastructure investment increasingly depends on debt, investors need to pay closer attention to interest expenses, refinancing requirements and future operating cash flow.

3. The Biggest Risk Is the Gap Between Spending and Returns

One of the questions I find most important is whether AI infrastructure spending can generate adequate returns on invested capital.

Demand for computing capacity is clearly growing, but that does not automatically mean every infrastructure investment will be profitable.

Data centres require substantial electricity, maintenance, networking equipment and ongoing hardware upgrades.

GPUs and other specialised equipment also depreciate, meaning companies must continue investing to remain competitive.

There is a possibility that some infrastructure built today could become less economically attractive before its original investment has been fully recovered.

This is particularly important for companies operating with high debt levels or relying on optimistic assumptions about future utilisation rates.

The AI opportunity is real, but the financial returns will not necessarily be distributed equally across the industry.

4. Higher Borrowing Could Create Wider Market Risks

The bond market is worth watching because the consequences of excessive borrowing would not necessarily remain limited to technology companies.

If corporate debt issuance continues increasing, investors may demand higher yields to compensate for additional risk.

Higher borrowing costs could affect companies planning further infrastructure expansion. It could also put pressure on highly leveraged businesses that need to refinance existing obligations.

There is also a potential knock-on effect for equity markets.

If investors become concerned about debt sustainability, they may reassess the valuations of companies whose future growth depends heavily on continued AI spending.

However, a busy bond issuance calendar alone does not indicate an impending credit crisis. The more important indicators are credit spreads, debt maturity profiles, interest coverage and the ability of companies to generate sufficient cash flow.

5. Not Every AI Company Will Be a Winner

I think this is where investors need to become more selective.

The first phase of the AI boom rewarded companies supplying the essential infrastructure, particularly semiconductor manufacturers and major cloud providers.

The next phase may increasingly distinguish between companies that can monetise AI successfully and those that are simply spending heavily to remain competitive.

For example:

• Semiconductor companies benefit from demand for advanced computing hardware, but must manage cyclical demand and customer concentration.

• Cloud providers can generate recurring revenue from AI workloads, but face enormous infrastructure and capital expenditure requirements.

• AI software companies have opportunities to generate higher-margin revenue, provided customers are willing to pay for measurable productivity improvements.

• Data-centre operators may benefit from long-term capacity demand, but their returns depend heavily on financing costs, utilisation and power availability.

The winners will not necessarily be the companies spending the most money. They will be the companies that can turn that spending into sustainable free cash flow.

My Investment Takeaway

I am not suggesting that the AI boom is about to collapse. The underlying demand for computing power, cloud services and AI applications remains an important long-term growth opportunity.

However, I think the market is moving into a stage where balance sheets and cash flow deserve as much attention as revenue growth.

A company can report impressive revenue growth while simultaneously consuming enormous amounts of cash to build the infrastructure needed to support that growth.

For investors, I would focus on three measures:

1. Free cash flow: Is the company generating enough cash to fund its expansion, or becoming increasingly dependent on external financing?

2. Debt and interest costs: Is borrowing growing at a manageable pace relative to operating earnings and future cash generation?

3. Return on invested capital: Are new data centres and AI infrastructure generating returns that justify their enormous upfront costs?

I also think the bond market deserves closer attention. If borrowing costs rise or credit spreads widen, the impact could extend well beyond the companies directly involved in AI infrastructure.

The AI revolution may create enormous wealth, but building it is not free. Ultimately, shareholders will benefit only if the revenue and profits generated by these investments justify the capital committed today.

For discussion purposes only. Not financial advice.