The AI investment boom is becoming a financial stability issue
AI investment could reach trillions of dollars by 2030, and companies are increasingly turning to bonds, banks and private credit to fund it. The RBA says that is creating a new set of financial-system risks.
The global artificial-intelligence investment boom is beginning to change the structure of financial markets.
The Reserve Bank's latest Financial Stability Review estimates that as much as US$7.7 trillion of AI-related capital expenditure could occur by 2030. Historically, technology companies relied heavily on retained earnings, equity and venture capital. But the scale of today's investment in data centres, semiconductors, computing equipment and electricity infrastructure is increasingly forcing companies into debt markets.
Some estimates cited by the RBA suggest more than one-third of planned AI capital expenditure could ultimately be debt financed.
The shift is already visible. Technology-related companies accounted for nearly 30% of US investment-grade corporate bond issuance in the first half of 2026. Funding is also increasingly coming from bank loans, private credit and asset-backed securities.
Why could this become a financial-stability problem?
Debt changes who bears the risk.
If an equity-funded AI investment disappoints, shareholders principally absorb the loss. As borrowing increases, banks, bondholders, private-credit funds and institutional investors become increasingly exposed.
The RBA identifies several vulnerabilities.
First is overinvestment. Data-centre construction is expanding at exceptional speed on assumptions that AI demand and revenues continue growing rapidly. If adoption is slower than anticipated, some projects may not generate enough income to service their debt.
Second is a mismatch between debt maturity and technology life. Companies are issuing debt with maturities of 20 or 30 years to finance assets such as chips and cooling equipment that may become technologically obsolete much sooner.
Third is opacity. Increasingly, data centres are financed through separate special-purpose vehicles. The RBA cites estimates that hyperscalers' financial obligations associated with these off-balance-sheet structures may already total US$1–1.5 trillion.
Finally, financing is becoming circular. Cloud providers can invest in AI developers that then purchase their computing services; chipmakers can finance customers that subsequently buy their chips.
Australia is not immune. The RBA says domestic data-centre operators have relied substantially on debt, including syndicated loans involving Australian and foreign banks. Current domestic exposures remain relatively small, and lenders appear cautious.
The immediate risk is therefore predominantly global.
But the underlying lesson is familiar from previous investment booms: technological transformation can be economically valuable while still producing financial losses if too much capital is committed too quickly.
|
|||||||||||||||||