Big Tech built its reputation on mountains of cash. It is building its AI empire on debt, and the bill is starting to land in Europe. The five biggest builders of AI data centres in the US — Alphabet, Amazon, Meta, Microsoft, and Oracle — have collectively doubled their debt over the past five years, amassing approximately $350bn according to data compiled by Bloomberg. This massive borrowing binge is a bet that cutting-edge artificial intelligence will eventually generate enough revenue to pay it all back.

For now, the interest payments remain a rounding error relative to their earnings. The five companies paid a combined $10bn on their debt last year, more than double the $4.5bn they paid in 2019. Yet that figure is tiny compared to Google’s quarterly cash flow of roughly $64bn. These are still some of the most profitable firms on Earth. Microsoft’s operating income surpassed $100bn last fiscal year, and Amazon’s annual revenue exceeded $574bn. But the strain is starting to show at the edges.

Amazon’s free cash flow turned negative in the March quarter, a stark shift from the $15bn surplus it reported a year earlier. Oracle’s debt reached about 2.5 times its sales, and S&P Global Ratings downgraded its credit rating to one notch above junk status on Thursday, directly citing the company’s aggressive AI spending. Meta, which spent $27bn on capital expenditures in 2024, most of it for AI infrastructure, has seen its debt grow by $20bn since 2020.

The market is starting to flinch

Investors have historically lapped up tech bonds, but that appetite may be waning. Amazon’s $25bn bond sale this week met a notably cool reception, the softest launch for a hyperscaler bond since Meta’s offering last October. Traders are now dumping older tech bonds, including those from Amazon, Nvidia, and Oracle, simply to make room for the new flood of issuance. The buyers are running out of space, not confidence, but the warning signs are stacking up.

“Credit risk is too undervalued right now in the market,” Vishal Khanduja, a fixed-income portfolio manager at Morgan Stanley, told Bloomberg TV. The yield spreads on investment-grade corporate bonds have compressed to near-historic lows, meaning investors are not being compensated for the increasing leverage. If the AI revenue boom falters, these companies could face a refinancing crunch.

The broader bond market is also showing fatigue. The total corporate bond issuance in the US reached a record $1.5 trillion in 2024, and tech companies accounted for a disproportionate share. With interest rates still elevated—the Fed’s benchmark rate sits at 5.25%—the cost of servicing that debt is rising. Every quarter, the interest expense eats into profits that could otherwise fund share buybacks or dividends.

Why this reaches Europe

Here is the part that should matter on this side of the Atlantic. American tech has run short of dollars to borrow, so it has turned to Europe. Hyperscalers issued virtually no non-dollar bonds in 2024. By 2026, such multi-currency borrowing is expected to become a core part of their funding strategy.

Morgan Stanley predicts that US Big Tech’s euro-denominated borrowing could reach €50bn this year, as reported by TechFundingNews. That would make them the single largest source of corporate debt in the eurozone, surpassing even the French government’s issuance. Alphabet alone has borrowed in yen, Canadian dollars, Swiss francs, and sterling within the past 12 months, and even sold a 100-year bond in euros—a maturing debt that will outlast most of its current workforce.

When American giants crowd into the same euro debt that European scale-ups and infrastructure funds rely on, the cost of money shifts here too. A startup in Munich or Paris with no AI exposure at all can end up paying more, simply because Amazon got there first. The spread between investment-grade and high-yield bonds in Europe has already widened by 30 basis points since the start of the year, partly due to this crowding effect.

European institutional investors, such as pension funds and insurance companies, have traditionally been large buyers of corporate bonds. But as US tech issuers increase their supply, these buyers are forced to allocate more of their portfolios to lower-yielding securities, pushing up yields for other borrowers. The European Central Bank’s rate hikes have already tightened financial conditions; the additional supply from Big Tech could exacerbate the situation.

The Intel warning

Not everyone is convinced the debt is sustainable. “It seems like a lot of demand hype that is very aspirational at this point,” said Jason Pompeii, a director at Fitch Ratings. The cautionary tale sits just one industry over. Intel spent years loading up on debt, missed the AI chip boom entirely, and needed a US government bailout under the CHIPS Act plus a $5bn investment from Nvidia to survive. The chipmaker’s debt-to-EBITDA ratio soared to over 4x before it finally began restructuring.

Intel’s experience highlights the risk of betting on a technology that may not mature as quickly as expected. AI models require massive computing power, but monetization remains elusive. Many enterprises have yet to see a return on their AI investments; a McKinsey survey found that only 10% of companies have achieved significant revenue gains from AI. If the adoption curve flattens, the billions spent on data centers may never be recouped.

On the other hand, Meta CEO Mark Zuckerberg insists that demand for computing power keeps outstripping supply. “We’re building for the long term,” he said on the most recent earnings call. Amazon’s Andy Jassy echoed that sentiment, stating he has “high confidence this will be monetized.” Gil Luria, an analyst at DA Davidson, agrees the load looks manageable for now: “If they were borrowing an order of magnitude more? That would look bad.”

The growing debt pile in historical context

The AI build-out has quietly become one of the largest debt bets in corporate history. The combined $350bn debt of the five hyperscalers is larger than the entire corporate bond market of South Korea. One research shop, Autonomous Research, estimates that the AI debt market could reach $7tn by 2029 if current trends continue.

To put that in perspective, the dot-com bubble saw tech companies pile on about $500bn in debt (in today’s dollars) at its peak—but those were mostly high-yield junk bonds. Today’s debt is predominantly investment-grade, but the concentration risk is higher. Alphabet, Amazon, Microsoft, Meta, and Oracle now account for 12% of the entire US investment-grade bond index, up from 4% five years ago.

The share buybacks that once defined these companies have all but stopped. Amazon repurchased $60bn of its own stock in 2022; last year that figure fell to $10bn. Microsoft, which spent $28bn on buybacks in fiscal 2023, reduced that to $18bn in 2024. The cash that would have gone to shareholders is instead being funneled into AI infrastructure. Investors are effectively being asked to wait for a payoff that may take years.

The gamble is no longer just whether AI works. It is whether the revenue arrives before the debt does. That same wager is driving Nvidia’s bond sales and ByteDance’s borrowing. Nvidia, the dominant supplier of AI chips, has issued $12bn in bonds this year to fund its own expansion, while ByteDance borrowed $5bn to build data centers for TikTok’s AI features.

What this means for European tech

For European startups, the ripple effects are already visible. The average yield on euro-denominated corporate bonds rated BBB has risen by 15 basis points since January, even as the ECB has started cutting rates. Smaller borrowers in the technology sector are having to pay premiums of 50 to 100 basis points more than they would have a year ago, according to data from Dealogic.

Infrastructure funds that finance renewable energy and digital connectivity projects are also feeling the pinch. The European Investment Bank has warned that crowding out could delay the green transition, as clean-tech ventures compete with Big Tech for scarce capital. Meanwhile, European cloud providers like OVHcloud and Deutsche Telekom’s T-Systems face higher financing costs, making it harder to compete with AWS and Azure.

The irony is that European regulators have been pushing for more digital sovereignty and homegrown AI capabilities. But if the cost of capital becomes prohibitive, that ambition may stall. The European Commission’s €7.5bn Digital Europe Programme is a drop in the ocean compared to the $150bn that US hyperscalers are spending annually on data centers.

In the end, Europe just became a place where that wager gets settled. As US tech giants borrow in euros, they are not just funding their own AI dreams—they are reshaping the continent’s credit markets, whether Europeans like it or not.

Source: TNW | Amazon News