To The Moon

The AI investment boom is creating enormous opportunities for semiconductor manufacturers, cloud providers and data-centre operators. But there is another side to this expansion that deserves more attention: how are these companies financing it, and what happens if the expected returns take longer to arrive?

On Monday, Oracle led a group of 18 companies issuing bonds in the US high-grade corporate bond market. This was the largest number of issuers entering the market in a single day since September 8. Dealers also expected total US investment-grade bond issuance to reach approximately $45 billion for the week.

The timing, ahead of the US midterm elections and Friday’s non-farm payrolls report, is notable. Companies may be taking advantage of a favourable financing window before potential market volatility.

However, I think the more important issue is what this borrowing activity tells us about the financial demands of the AI infrastructure boom.

1. AI infrastructure requires enormous amounts of capital

Building AI infrastructure is expensive. Data centres require specialised buildings, advanced chips, networking equipment, cooling systems and substantial electricity capacity.

The investment does not stop once a facility is built. Companies must continue spending to upgrade computing equipment, expand capacity and maintain infrastructure as technology evolves.

For major technology companies, strong operating cash flow can help fund these investments. However, when capital expenditure grows faster than internally generated cash, borrowing becomes another way to finance expansion.

This is not necessarily a problem. Debt can be a sensible financing tool when investments generate returns that exceed their funding costs.

The risk emerges when companies commit to increasingly large projects without sufficient visibility on future customer demand and profitability.

2. Negative free cash flow deserves closer attention

Free cash flow is one of the most useful measures when assessing whether a company’s investment programme is financially sustainable.

A company can report strong revenue growth and still experience pressure on free cash flow if capital expenditure rises significantly.

This is particularly relevant for businesses expanding AI data-centre capacity. Spending happens upfront, while the revenue generated by new infrastructure may take years to fully develop.

Investors should distinguish between temporary negative free cash flow caused by strategic expansion and a more persistent inability to generate sufficient cash.

I would monitor:

• Operating cash flow: Is the core business generating enough cash?

• Capital expenditure: How quickly is infrastructure spending increasing?

• Free cash flow: Is cash generation improving or deteriorating?

• Debt levels: Is borrowing growing faster than the company’s ability to service it?

A company investing heavily today may eventually generate substantial returns. But investors need evidence that the investment is translating into paying customers, recurring revenue and stronger margins.

3. Oracle deserves particular attention

Oracle is an interesting company to watch because its AI opportunity is closely connected to demand for cloud infrastructure and data-centre capacity.

Expanding capacity could position Oracle to capture more business from customers requiring computing resources for AI workloads.

However, building that capacity requires significant funding. The financial outcome depends on whether customer demand, utilisation rates and pricing can support the cost of the infrastructure.

The key question is not simply how much capacity Oracle can build. It is how quickly that capacity becomes productive and generates sufficient cash returns.

Investors should pay attention to debt growth, financing costs, contracted customer commitments and the timing of expected revenue.

A large order book can provide visibility, but it does not automatically guarantee attractive returns.

4. The credit market may be telling us something

The growing number of companies accessing the bond market raises another important question: how much additional debt can the technology sector absorb?

Investment-grade companies generally have stronger credit profiles than speculative-grade borrowers. Their ability to issue bonds does not automatically indicate financial distress.

In fact, companies may borrow because they have access to attractive financing terms and want to secure funding ahead of uncertainty.

Nevertheless, investors should watch whether borrowing costs increase, credit spreads widen or lenders become more selective.

If companies need to pay increasingly high interest rates to finance infrastructure projects, the economics of those investments become less attractive.

There is also a potential issue of concentration. If many large companies borrow heavily to build similar infrastructure at the same time, the sector could become more exposed to a slowdown in AI spending or a change in customer demand.

5. What happens if AI returns take longer than expected?

This is probably the biggest risk worth considering.

The current AI investment cycle is built on expectations of sustained demand for computing power. If AI adoption continues accelerating, the infrastructure being built today could generate substantial long-term revenue.

But there is another possible outcome.

If customers become more cautious about AI spending, infrastructure utilisation could fall short of expectations. Companies may then face higher depreciation costs, ongoing interest expenses and large capital commitments without generating the anticipated returns.

This could put pressure on earnings and free cash flow even if overall revenue continues growing.

The risk is not necessarily that AI demand disappears. It is that infrastructure spending gets ahead of the commercial returns.

6. How can investors manage the risk?

I would avoid treating every bond issue as a warning sign. Borrowing is a normal part of corporate finance, particularly for companies undertaking large capital investment programmes.

Instead, investors should look at the quality of the balance sheet and the returns generated by that borrowing.

Some important measures include:

• Net debt to EBITDA: Does the company’s debt remain manageable relative to operating earnings?

• Interest coverage: Can operating profits comfortably cover interest expenses?

• Free cash flow trends: Is the business moving towards stronger cash generation after capital expenditure?

• Debt maturity schedule: When will existing borrowing need to be refinanced?

• Return on invested capital: Are new infrastructure investments generating returns that justify their cost?

I would also be careful about concentrating a portfolio entirely in companies benefiting from AI infrastructure spending. Different businesses may be exposed to the same underlying risk, even if they operate in different parts of the supply chain.

My investment view

I remain interested in the long-term AI opportunity, but I think the market needs to pay more attention to how that opportunity is being financed.

Revenue growth and infrastructure demand are important, but they do not tell the whole story. Companies also need to demonstrate that their investments can generate sustainable free cash flow and returns that justify the debt and capital commitments involved.

The next phase of the AI investment cycle may separate companies that can turn infrastructure spending into profitable growth from those that simply keep spending to maintain their competitive position.

For me, the biggest warning sign would not be a single large bond issue. It would be a combination of rising debt, weakening free cash flow, increasing financing costs and disappointing returns on new infrastructure.

AI may be creating a massive investment opportunity, but the cost of building that future will matter just as much as the revenue it eventually generates.

For discussion purposes only. Not financial advice.