How AI giants could raise the cost of borrowing in Europe

AI giants are borrowing on an enormous scale to fund data centres and other AI infrastructure. Their bonds compete with bonds issued by European governments, companies and public lenders for investors’ money. When demand for borrowing is greater than the available capital, borrowers may need to offer higher interest rates.

France shows why this matters. Its 10-year bond yield rose from about 3.7% to nearly 5% in six months. The government’s interest bill is expected to rise from €79 billion to €91 billion next year, then above €100 billion by 2028. New AI borrowing could add to that pressure.

European companies and regional authorities face the same competition. Higher interest costs can reduce earnings and make it harder to finance factories, equipment, research, schools, transport and infrastructure. The ECB warned that concentrated AI debt could spill over into sovereign and supranational borrowing markets.

Four of the Big Five hyperscalers—Amazon, Alphabet, Meta and Oracle—issued around $250 billion in dollar bonds from the start of 2025. They also sold a further €37 billion in euro-denominated bonds. These figures show how quickly AI financing has expanded in major bond markets.

Microsoft, previously the only Big Five company holding back, joined the borrowing spree with a $25 billion dollar bond sale. That brings the reported dollar issuance from these five companies to roughly $275 billion, although the article presents the currencies separately.

The borrowing is expected to grow. Goldman Sachs projects $420 billion in new AI hyperscaler bond sales next year. ING estimates about €60 billion of that issuance will be in euros. The larger flow could intensify competition for investors’ capital across Europe.

A bond is a tradable loan. An investor provides money to a company or government, and the issuer promises regular interest payments plus repayment of the principal at an agreed date. The article describes fixed-income funds as investors that mostly buy bonds because those securities provide regular interest.

Companies issue bonds to obtain money for major projects, such as AI infrastructure, plants, equipment or research. Governments and regions issue them to fund public services and infrastructure, including schools and transport. Bonds let borrowers raise substantial sums from many investors at once.

The article’s AI borrowing boom shows the scale of this method. Amazon, Alphabet, Meta and Oracle issued about $250 billion in dollar bonds since early 2025, plus €37 billion in euro bonds. Microsoft added a $25 billion sale. The borrowing creates competition for investors’ available capital.

Bond prices and yields generally move in opposite directions. When investors compete to buy a bond, its price rises and the return, or yield, falls. When many borrowers compete for investors’ money, issuers may have to offer more interest to attract buyers. That higher yield becomes a higher borrowing cost.

AI hyperscalers add unusually large supplies of new debt. BNP Paribas strategist Viktor Hjort called this “bond abundance” and said hyperscalers already create competition within the corporate sector and between companies and governments. The same investment funds must decide which bonds to hold.

The result is not that every bond rises by the same amount. Riskier or less attractive borrowers may need to pay more, especially when markets are already unsettled. The ECB warned that concentrated AI debt could spill over into sovereign and supranational bond markets, including EU and European Investment Bank borrowing.

These borrowers are vulnerable because many already face heavy fiscal or financial pressure. France has public debt equal to 119% of GDP and a deficit above 5%. Its interest bill is projected to rise sharply. Germany and Italy are also named among countries competing for the same investment capital.

European businesses are under pressure too. Goldman Sachs found that interest costs as a share of sales at Stoxx 600 companies rose by nearly half since 2022, cutting earnings by more than 5%. Cities and regions borrow to fund schools, transport and infrastructure, so higher yields can affect everyday public investment.

The timing makes the competition more serious. Energy prices linked to the Iran war and France’s political crisis recently pushed bond yields higher. AI hyperscalers plan much more borrowing next year. That extra supply could increase pressure on already stretched national, corporate and local borrowers.

Governments facing higher yields must devote more revenue to interest rather than public services or investment. France’s Finance Ministry may need tens of billions of euros in extra cuts, tax increases or borrowing by the end of the decade. The article also warns that the EU and European Investment Bank could face higher financing costs.

Companies may respond by protecting debt payments. European firms could cut investment in plants, equipment and research, and reduce shareholder payouts. Goldman Sachs already found that rising interest costs had lowered Stoxx 600 earnings by more than 5% since 2022. Higher rates would intensify that trade-off.

Local governments could face similar choices. Cities and regions borrow for schools, transport and infrastructure, so costlier debt could delay or shrink projects. The article does not predict specific cuts, but it identifies debt repayment, lower investment and higher taxes as possible consequences.

Government bonds often attract investors when uncertainty rises because nervous investors seek a safer place for their money. Greater demand pushes bond prices up and yields down, reducing governments’ borrowing costs. This can cushion public finances when markets are unsettled.

The article says that normal pattern may be weaker here. AI hyperscalers are issuing huge amounts of debt at the same time that governments and companies need capital. Instead of investors moving mainly into government bonds, they may divide funds among competing corporate and sovereign borrowers. That can keep yields elevated.

Recent market movements illustrate the concern. Energy prices linked to the Iran war and France’s political crisis drove yields higher, while government and company borrowing costs rose together. With Goldman Sachs expecting $420 billion in new AI hyperscaler bonds next year, this competition could become more pronounced.