The artificial intelligence investment boom is entering a new phase. Technology companies are committing unprecedented amounts of capital to data centres, advanced semiconductors, networking equipment and computing infrastructure. The scale of this spending reflects expectations that AI will transform productivity and create significant future revenues.
However, a fundamental corporate-finance question is becoming increasingly important: are companies generating enough cash today to support the investment they are making for tomorrow?
The answer is mixed. Major hyperscalers continue to generate substantial earnings and operating cash flows, but the scale and speed of their capital expenditure are creating increasing pressure on free cash flow. This does not necessarily indicate financial weakness. Instead, it highlights the extent to which the AI investment cycle depends on future earnings growth to justify today’s spending.
AI Capital Expenditure Is Rising Rapidly
Capital expenditure, or capex, represents spending on long-term assets such as data centres, servers and other infrastructure. For AI companies, these investments are particularly important because developing and operating advanced AI systems requires enormous computing capacity.
The five largest hyperscalers are expected to spend more than $1 trillion on AI-related capital expenditure during 2025 and 2026, according to the BIS. Their investment commitments are growing faster than the earnings and free cash flow generated by some of these companies, increasing the need for external financing.
The scale of the spending is also expected to remain high. The IMF estimates that major hyperscalers could face around $3.4 trillion of AI-related capital expenditure through 2029.
This creates an unusual corporate-finance situation: companies with historically strong cash generation are committing capital at a pace that could increasingly exceed internally generated funds.
Earnings Have So Far Supported Much of the Investment
It is important not to interpret rising capex as evidence that AI companies are already financially overstretched. The IMF’s analysis shows that, for major hyperscalers, earnings growth has so far broadly kept pace with capital expenditure. Free cash flow has also remained at relatively high levels. This distinguishes the largest technology companies from many other businesses participating in the broader AI investment cycle.
This distinction matters.
A company can spend heavily on infrastructure without creating a financial problem if its operating cash flows and balance sheet are sufficiently strong. High investment can therefore be a sign of confidence in future growth rather than financial distress.
The more important question is what happens if capital expenditure continues increasing faster than earnings.
The Pressure on Free Cash Flow
Free cash flow is particularly important in assessing the sustainability of the AI investment cycle.
A company can report strong accounting profits while simultaneously experiencing pressure on free cash flow because large amounts of cash are being invested in property, equipment and infrastructure.
The BIS has highlighted this issue, noting that AI-related investment is testing the ability of companies to finance expansion through operating cash flows alone. Companies that historically relied heavily on their own cash generation are increasingly turning towards debt as investment requirements rise.
This creates a critical relationship:
AI investment → higher capex → lower near-term free cash flow → greater dependence on future earnings growth
The model can work extremely well if AI revenues and productivity gains develop as expected. But if monetisation takes longer than anticipated, companies may face a prolonged period in which capital expenditure remains high while returns arrive more slowly.
Also read: How Cash Flow Management Reveals the True Financial Health of a Business
Why Companies Are Willing to Invest Ahead of Earnings
There is a rational reason for companies to invest before the full economic benefits become visible.
AI infrastructure has characteristics of a competitive investment race. Companies may believe that securing computing capacity today is necessary to remain competitive in the future.
The BIS describes the current AI build-out as an investment race in which firms may have incentives to commit substantial resources because the eventual market could be dominated by a relatively small number of successful players. This can create a situation in which individual companies continue increasing investment even when the immediate financial return is uncertain.
From a strategic perspective, under-investing may carry its own risk. A technology company that waits for clear evidence of AI demand before expanding infrastructure could potentially find itself behind competitors that invested earlier.
Consequently, management teams are effectively making a trade-off between near-term cash-flow efficiency and long-term competitive positioning.
The Financing Gap Could Become More Important
The most significant concern is not simply that companies are spending heavily. It is that sustained investment could eventually create a gap between cash generated by operations and cash required for investment.
The IMF estimates that future earnings and cash buffers could become insufficient to fund the projected level of AI capital expenditure for major hyperscalers, potentially creating balance-sheet pressures in coming years.
The financing response is already visible.
The BIS reports that hyperscalers’ gross corporate bond issuance exceeded $100 billion in 2025, with much of the borrowing structured at longer maturities. This represents an important shift for companies that historically relied heavily on internal cash generation.
Debt financing does not automatically make an investment cycle dangerous. However, it changes the financial equation because future cash flows must ultimately support both investment and debt obligations.
Private Credit Is Adding Another Layer
Traditional corporate bonds are not the only source of financing. AI infrastructure is increasingly being funded through private credit and off-balance-sheet structures. In some arrangements, separate entities finance data centres using debt from private investors, while technology companies enter into long-term leases or capacity agreements.
The BIS describes these arrangements as forms of “shadow borrowing” because the economic obligations can resemble debt even when the associated borrowing does not appear directly on the hyperscaler’s balance sheet.
This matters because assessing AI investment purely through reported corporate debt may not provide the complete picture of the financial commitments supporting the infrastructure boom. The greater the reliance on external financing, the more important future cash flows become.
The Real Test Is AI Monetisation
Ultimately, the sustainability of AI spending depends on whether investment produces sufficient economic returns. Companies are investing today in the expectation that AI will generate higher revenues, reduce operating costs, improve productivity or create entirely new businesses.
The IMF has emphasised that AI has the potential to produce substantial productivity gains, but the timing and scale of those gains remain uncertain. The broader economic benefits of general-purpose technologies can also take years to diffuse through businesses and industries.
This creates an important distinction between investment growth and investment returns.
Rising capex demonstrates that companies are committing resources to AI. It does not, by itself, demonstrate that those investments will generate sufficiently high returns.
For investors, the focus therefore needs to shift from simply tracking how much companies spend on AI towards evaluating whether the spending is producing incremental revenues, margins and cash flows.
What Investors Should Watch
Several financial indicators can help investors assess whether AI companies are investing ahead of earnings:
1. Capex growth versus operating cash flow
If capital expenditure continues growing faster than operating cash flow, internally generated funding becomes less sufficient.
2. Free cash flow
A sustained deterioration in free cash flow can indicate that rising investment is consuming an increasing share of internally generated cash.
3. Revenue generated by AI products
The critical question is whether AI-related revenues are beginning to grow quickly enough to justify the infrastructure investment.
4. Return on invested capital
Investors should consider whether incremental investment is generating attractive returns rather than simply increasing the absolute scale of the business.
5. Debt issuance
Growing reliance on corporate bonds, leveraged loans or private credit would indicate that investment requirements are increasingly exceeding internal financing capacity.
6. Data-centre utilisation
Infrastructure investment ultimately needs to translate into productive use. Underutilised capacity could reduce the returns generated by large capital commitments.
Also read: Capital Structure Explained: How Debt vs. Equity Drives Business Growth
Not All AI Spending Represents the Same Risk
The AI investment cycle should not be treated as a single homogeneous phenomenon. The largest hyperscalers generally possess substantial earnings, cash balances and access to investment-grade debt markets. Their financial position is therefore different from smaller companies attempting to finance AI expansion primarily through external borrowing.
The IMF notes that demand for hyperscaler debt remains healthy, even as the sector prepares for very large future capital expenditure.
At the same time, the BIS has warned that the scale and speed of the investment boom create vulnerabilities if expected AI returns fail to materialise. The distinction is therefore between investing ahead of earnings and investing without a credible path to earnings.
The former can be a rational growth strategy. The latter can become a financial problem.
Conclusion
The AI investment boom demonstrates an important principle of corporate finance: cash flows matter as much as earnings when companies undertake enormous capital programmes.
Major technology companies continue to generate substantial profits and operating cash flows, but the scale of AI-related capital expenditure is increasing rapidly. If investment remains ahead of internally generated cash, companies will increasingly need to rely on their balance sheets, bond markets and private credit.
That does not mean the AI investment cycle is unsustainable. It means that its long-term success increasingly depends on the ability of AI-related revenues and productivity gains to catch up with today’s spending.
For investors, the key question is therefore not simply “How much are companies spending on AI?” It is “Will the future cash flows generated by that investment be large enough to justify the capital being committed today?” That is likely to be one of the most important financial questions of the AI investment cycle.
Also read: Why AI Could Reshape Corporate Bond Markets