
Central banks and analysts warn of overvaluation at leading AI companies
This article was created with the help of artificial intelligence.
Key Takeaways
- The ten largest stocks in the S&P 500 now account for 35 percent of the overall index, more than the 25 percent at the peak of the dot-com bubble in 2000.
- Global investment in artificial intelligence is expected to exceed 2.5 trillion US dollars in 2026, an increase of 44 percent over the previous year.
- Leading AI stocks such as NVIDIA and Broadcom have price-to-sales ratios above 30, a level that historically preceded sharp market corrections.
- At least one bank that had arranged loans for data centers sold parts of the debt at a price below 90 cents per dollar.
- The International Monetary Fund warned in October 2026 that a bursting of the AI bubble could shake global financial markets.
Capital markets show a contradictory picture in October 2026: on the one hand, 88 percent of companies surveyed worldwide use artificial intelligence, and 70 percent deploy generative AI in at least one function, according to data from Stanford University. On the other hand, central banks and analysts warn of overvaluation in some segments of the industry. The Bank for International Settlements (BIS) found in its 2026 Annual Economic Report that the implied long-term earnings growth of leading AI companies is significantly above historical benchmarks, while risk premiums have simultaneously declined.
Concentration reaches record level
The ten largest stocks in the S&P 500 now account for 35 percent of the overall index, more than at the peak of the dot-com bubble in 2000, when the figure stood at 25 percent. This concentration raises the risk of a sector-specific correction that could weigh heavily on the overall market. Global investment in artificial intelligence is expected to exceed the 2.5 trillion US dollar mark in 2026, an increase of 44 percent year on year. Around half of this sum flows into data centers and infrastructure.
NVIDIA shares rose by more than 880 percent over the past three years. Leading AI stocks such as NVIDIA and Broadcom have price-to-sales ratios above 30, a level that historically preceded sharp market corrections. AI and semiconductor stocks reached valuations in 2024 and 2025 that were last seen during the dot-com era.
Private valuations and debt markets under pressure
OpenAI and Anthropic achieved exceptionally high valuations in the private capital market in 2026, which imply high expectations for future revenue growth and margins. At the same time, IPOs of AI companies are being postponed because investors are reluctant to accept the valuations being set.
Credit markets are sending warning signals: banks are increasingly showing discomfort with financing AI infrastructure. At least one bank that had arranged loans for data centers sold parts of the debt at a price below 90 cents on the dollar. Other banks struggled to place multi-billion-dollar loans for data centers in Texas and Wisconsin that were to be leased to large technology companies. The rising refinancing risk is increasingly drawing the attention of risk examiners to the financing of so-called neocloud providers.
Differences from established technology companies
Microsoft, Alphabet, Meta and Amazon finance their data centers from ongoing cash flow rather than through debt or capital increases, an important difference from the dot-com era. Smaller providers, by contrast, increasingly rely on debt-financed expansion, which concentrates bubble risks.
Three risk zones in focus
The main risks of an AI bubble focus on three areas: first, extreme valuations of privately held companies; second, debt-financed data center expansion; and third, long-term computing capacity contracts that presuppose very high future cash flows. This structure differs from the broad speculation of the dot-com era, when many companies had neither revenues nor viable business models.
Analysts recommend watching observable indicators: customer revenue growth, credit spreads, utilisation of graphics processors, free cash flow and contract cancellations can signal overheating or cooling at an early stage. A reliable prediction of the exact timing of a correction does not exist.
Systemic effects of a correction
In October 2026, the International Monetary Fund (IMF) warned that a bursting of the AI bubble could shake global financial markets. A loss of investor confidence would lead to delays and cuts in AI investment and weigh on gross domestic product; in an extreme case, a pronounced recession could follow. Financial markets are increasingly exposed to an AI-driven correction, particularly if the billions flowing into technology fail to deliver the expected earnings.
Options for investors
Investors should conduct thorough individual company analyses, diversify across the AI value chain and adjust position sizes to the inherent volatility. Exposure to high-quality AI beneficiaries can be combined with strict risk management. Corporate customers are advised to finance measurable workflows rather than speculative implementations, maintain model portability and avoid long-term infrastructure commitments not backed by validated demand.
The central analytical distinction is this: a bubble can form around useful technology, as overinvestment and lasting technological value are not mutually exclusive. The closest historical analogy is not the disappearance of AI like failed dot-com companies, but the possibility that too much capital flows into valuable infrastructure at unsustainable prices.
Sources
- AI Bubble 2026: Is It Bursting? Data & Warning Signs
- Is Your Portfolio at Risk and What to Do About It Now
- Is AI in a Bubble? Valuation Risks, Crash Signs and Safer Investing
- How An AI Bubble Burst Could Shake Global Financial Markets
- AI bubble - Wikipedia
- AI bubble could burst and take global economy with it, IMF warns
- 7 Reasons The AI Bubble Is About To Pop
- 2026 Market Analysis