Source: in-cyprus.philenews.com
The artificial intelligence boom is increasingly being financed by an unlikely source: the long-term savings of Europeans.
European pension funds and insurers are pouring money into long-term debt issued to finance the artificial intelligence build-out in the United States. The bonds are often considered among the safer parts of the corporate debt market. But regulators are beginning to ask whether investors are underestimating the risks hidden inside the AI spending spree.
Behind the race to build data centres, buy chips and expand electricity infrastructure is a rapidly growing market for debt. US technology companies are borrowing heavily to fund their AI ambitions, while European pension funds and insurers are among the institutional investors looking for long-term, relatively safe assets in which to put their money.
The result is a financial connection that is easy to miss.
A European worker puts money into a pension fund. The pension fund needs to invest that money for decades. A US technology company needs billions of dollars to build data centres. The company issues a long-term bond. The pension fund buys some of that debt.
European savings can therefore end up financing American AI infrastructure, and for the moment, much of that debt looks reassuringly safe.
The AI boom needs an extraordinary amount of money
The biggest technology companies, including Microsoft, Alphabet, Amazon, Meta and Oracle, are spending unprecedented amounts on AI infrastructure. The investment goes well beyond buying computer chips. It includes enormous data centres, power supplies, networking equipment and the physical infrastructure required to operate them.
The Bank of England says expectations for the future investment needs of the major AI companies have risen sharply. At the time of its December 2025 Financial Stability Report, estimates for their combined capital expenditure in 2028 were below $600 billion. By July 2026, the figure had risen above $1 trillion. That money has to come from somewhere. Some comes from the companies’ own cash flow. Some comes from issuing shares. But an increasingly important part is coming from debt.
According to the Bank of England, the five major AI hyperscalers accounted for more than 15% of new US investment-grade debt issuance by early May 2026, even though they represented only about 3% of the outstanding US investment-grade debt market at the end of 2025.
Why pension funds like this debt
For a pension fund, lending money to a highly rated technology company can look quite different from buying its shares. Shares are risky: if the company’s profits collapse, shareholders can lose a lot of money.
A bond is different. When an investor buys a bond, they are essentially lending money to the company. In return, the company promises to pay interest and eventually repay the principal.
And pension funds have a particular reason to like long-term bonds. They have liabilities that stretch decades into the future. They need assets that can generate relatively predictable returns over similarly long periods.
This makes a 10-, 15- or 20-year bond issued by a highly rated company potentially attractive. The technology company gets the money it needs to build its AI infrastructure. The pension fund gets a stream of interest payments. On paper, it can look like a fairly straightforward transaction.
But there is a risk
The concern is not that Microsoft or Amazon are suddenly going to default. The bigger question is whether the enormous investments being made today will generate enough economic returns in the future to justify their cost.
The AI industry is making a gigantic bet. Companies are building infrastructure today on the assumption that demand for AI computing will remain extremely strong for many years. But technology changes quickly.
AI models could become dramatically more efficient. Demand could grow more slowly than expected or competition could push prices down. It could be argued that companies could discover that the revenue generated by AI services does not justify the enormous cost of building the infrastructure.
The debt, however, remains. A company that borrowed $20 billion still owes the money even if the data centre it financed becomes less valuable than expected.
This is one reason regulators are paying increasing attention to the debt behind the AI boom. The European Central Bank (ECB) has also warned that the AI investment boom could create financial-stability risks if the enormous spending needed to build data centres and computing infrastructure is increasingly financed through debt. The concern is not that the major technology companies are currently poor-quality borrowers, but that the sheer scale and speed of borrowing could leave investors exposed if expectations for AI revenues or productivity fail to materialise.
Europe is therefore financing a bet it does not control
This is perhaps the most intriguing part of the story. Europe has huge pools of long-term savings, whilst the United States has the world’s most aggressive AI investment programme.
A European pension fund doesn’t need to believe that AI stocks are going to double in value. It doesn’t even need to buy an AI company’s shares. It can simply buy a bond. But that bond is ultimately part of the financing machinery supporting the AI build-out.
If AI delivers the enormous productivity gains its advocates expect, the arrangement could work beautifully. Technology companies generate enormous revenues, they then repay their debts and investors (in this case European pension funds) collect interest.
But if the economics disappoint, the consequences could be wider than a fall in technology stocks meaning that certain projects could struggle to refinance their debt and some of the capital ultimately at risk could belong to European institutions managing the savings of millions of workers.
In a letter dated 28 August, Andrew Bailey, the Financial Stability Board’s (FSB) chair stated: “Markets remain vulnerable to a potentially disorderly correction that could spread across borders, particularly given fragilities in sovereign debt markets (including elevated issuance, shortening maturities, and the increased use of leverage by some market participants).”
“Vulnerabilities in private credit (including levels of interconnectedness with other parts of the financial system, liquidity mismatch and opacity); as well as stretched asset valuations (particularly artificial intelligence-related investments),” he added.
That does not mean European pensions are suddenly in danger. The amounts invested in any one project are generally only a fraction of an institution’s overall portfolio, and the largest technology companies remain financially powerful borrowers.
The question investors now have to answer
The fundamental question is no longer simply whether artificial intelligence will transform the economy. Its focus is more on whether the extraordinary amount of money being spent to prepare for that transformation will generate returns large enough to justify the debt being accumulated along the way.
An unnamed source who works in the financial sector on EU policy told en.philenews that one main concern is that “the market is pricing a demand that doesn’t yet exist for massively leveraged investment in this AI buildup.”
“In order for this to payoff there needs to be uptake in AI, which is not there yet. All business needs to take up AI for this to happen. This could lead to a worst case scenario where jobs replaced by AI and then AI pays up. That will inevitably lead to unemployment creating another political headache,” he added.
For Europe, the question is becoming increasingly difficult to ignore. The European Central Bank has warned that the euro area financial sector is becoming more exposed to US companies at the forefront of AI development, and that a sharp repricing of those companies could spill over into the European financial system. In its May 2026 Financial Stability Review, the ECB also said that AI-related companies and infrastructure were increasingly turning to credit financing, warning that this could become a financial-stability concern if it translates into a significant increase in business debt. The central bank has highlighted particular risks for European insurers and pension funds, which could face losses if stress in private-credit, bond or equity markets spreads.
For European investors, then, the AI boom is no longer simply an American technology story. European institutions have significant exposure to US assets, while the companies driving the AI build-out are increasingly dependent on debt markets. The ECB’s concern is not that these investments are necessarily unsafe today, but that a combination of high valuations, rapidly growing debt and concentrated exposure to a small group of US technology companies could leave European markets vulnerable if expectations around AI growth change abruptly. As the ECB puts it, concentrated exposures across public and private markets could result in “non-linear correlated losses” if sentiment shifts.
For now, the money is flowing and technology companies are building. Pension funds and insurers are buying and credit-rating agencies still largely regard the debt as safe. The unanswered question is what happens if in the future those bonds are turn out to be worth less than everyone currently expects.
Read more: