Despite geopolitical tensions, rising energy prices and high public debt, the global economy is proving remarkably resilient. “But this resilience does not necessarily mean safety,” said Athanasios Vamvakidis, deputy director of the IMF’s Monetary and Capital Markets Department, at the presentation of the latest Global Financial Stability Report.  

Behind this robust performance, however, significant risks for financial markets are building. At its annual meeting in Bangkok, the International Monetary Fund warned of a dangerous combination of rising bond yields, high debt and the increasingly debt-financed boom around artificial intelligence.

What is especially problematic is that a setback in AI investment could hit not only technology stocks, but also put banks and other financial institutions under pressure via the bond markets.

Tech groups are taking on massive debt

Building new data centers and AI infrastructure requires enormous sums. While the major technology groups have so far largely been able to finance their investments from internal resources, they are now increasingly turning to the capital markets.

According to IMF calculations, nearly ten percent of all newly issued investment-grade corporate bonds worldwide in 2026 will be accounted for by just five companies: Alphabet, Amazon, Meta, Microsoft and Oracle. In 2024, their share had been just 0.5 percent.

The tech groups’ interest costs have risen accordingly. In the second quarter, they totaled almost six billion dollars, more than twice as much as a year earlier.

As a result, the companies are increasingly competing with sovereigns for the same capital. Rising demand for financing could push bond market yields even higher.

1.4 trillion dollars in mutual obligations

The IMF is particularly critical of the financial interconnections within the AI industry. Since 2023, the experts have identified 14 major transactions in which companies simultaneously act as investors, lenders, suppliers and customers. These arrangements include equity investments of around 300 billion dollars and debt financing of 650 billion dollars.

In return, the financed companies committed to purchasing cloud services or AI chips worth a total of 1.4 trillion dollars in the future.

One example of such structures are investments by chip manufacturers in data center operators, who then use the capital they receive to buy chips from the investors. The problem: financing and future revenue are concentrated among a few companies. If one of these players runs into trouble, it could trigger a chain reaction.

IMF economist Jason Wu warns of a threefold risk: equity stakes lose value, loans may no longer be serviceable, and the expected revenue fails to materialize.

“Weaker AI investment or lower profitability could be a shock that triggers a market correction,” the IMF warns.

New borrowing is rising again

At the same time, conditions in bond markets are deteriorating. Yields on long-term government bonds have risen sharply in numerous industrialized countries. The rise in sovereign bond yields is “clearly” the biggest change in the risk picture since April, Athanasios Vamvakidis told Bloomberg. Alongside high budget deficits, developments in energy prices and rising inflation expectations are adding further pressure.

The IMF expects net new borrowing by industrialized countries to rise to 5.2 percent of gross domestic product in 2026. That would be the highest level since 2021.

The deficits are particularly pronounced in the US. According to the IMF, general government net borrowing there is likely to reach 8.5 percent of gross domestic product, up from 7.1 percent the previous year. Global government debt was already close to 94 percent of economic output in 2025.

How clearly investors are now distinguishing between different sovereign borrowers can be seen in Europe. The yield spread between ten-year French government bonds and German Bunds has recently risen to more than 140 basis points. The backdrop is France’s high public debt, persistent budget deficits and political uncertainty.

IMF stress test: banks need 68 billion dollars

The potential consequences of a simultaneous slump in AI valuations and a further rise in interest rates are shown by an IMF stress test involving 665 banks from 29 countries. In the simulated crisis scenario, equities lose more than a quarter of their value over three years. At the same time, yields on long-term bonds rise by three percentage points, while oil and other commodities become more than 50 percent more expensive.

Under these assumptions, the global banking system would remain adequately capitalized overall. However, institutions representing 22 percent of global banking assets would fall below a hard common equity Tier 1 ratio of seven percent. In the previous year’s test, this figure had still stood at 18 percent.

Even more serious: banks accounting for seven percent of global assets would even fall below the regulatory minimum ratio of 4.5 percent.

According to IMF calculations, these institutions would need a total of 68 billion dollars in additional equity capital. Banks in Asia would be particularly affected, followed by European institutions. This is explicitly a hypothetical crisis scenario and not a forecast.

Cyberattacks as an additional risk

Beyond the financial risks, the IMF also sees a growing threat from cyberattacks. In more than 70 percent of reported cases, software providers in 2026 had virtually no time to close a security vulnerability before attackers exploited it; five years ago, according to the IMF, it was still nearly 300 days.

A major attack on a large bank, payment service provider or cloud provider could therefore have far-reaching consequences. The increasing interconnection of the financial system in particular heightens the risk that disruptions spread to numerous institutions.

Global economy continues to grow nonetheless

Despite the warnings, the IMF does not expect an immediate slump in the global economy. Among the key supports, it says, are still the multi-billion-dollar investments in artificial intelligence.

For 2026, economists have slightly raised their global growth forecast to 3.1 percent. For next year, the IMF now even expects growth of 3.4 percent.

The US is expected to grow by 2.3 percent in both years. For the eurozone, the IMF forecasts 1.1 percent in the current year and 1.2 percent next year. Germany has received a significantly better forecast: instead of the previous 0.7 percent, the Fund now expects growth of 1.3 percent in 2026, followed by 1.1 percent a year later.

For Switzerland, the IMF expects real growth of 1.9 percent in the current year and only 1.2 percent for 2027.