Financial markets / AI bubble
Whether AI valuations form a bubble is possible but not proven; whether and when it bursts cannot be reliably predicted. Check your concentration risk: around one quarter (25 %) of the global MSCI World index is in a few US tech stocks. Avoid panic selling, maintain your time horizon – this is not investment advice.
Prices around artificial intelligence have risen sharply in just a few years. Whether this is a speculative bubble is a possibility – not a fact; whether and when it bursts cannot be predicted seriously. Several supervisory authorities (IMF, Bank of England, ESMA, BIS, US Federal Reserve) nonetheless warn of stretched valuations and market concentration.
The provision risk is less a single price drop than the cluster risk: around a quarter of the global MSCI World index is made up of the seven largest US tech stocks. Many pillar-3a, pension-fund and ETF products with a global reference are thus more concentrated in a few names than people think.
This hazard is deliberately kept sober: it is about preparedness education, not investment advice. A detailed, sourced analysis with the web of companies and price charts is in the Finance chapter.
Risk rating
Likelihood: Medium · Impact: Medium → Overall risk: Medium.
| Impact | |||
|---|---|---|---|
| Low | Medium | High | |
| High | |||
| Medium | This hazard's rating | ||
| Low | |||
Legend: Low (green) · Medium (yellow) · High (red). Overall risk is likelihood multiplied by impact.
A strong concentration in a few AI stocks can also hit broad "world" investments – a cluster risk for pillar 3a, pension funds and ETFs.
Immediate measures
- Review your own cluster risk in 3a/pension fund/ETFs
- No panic selling – keep your time horizon and strategy
- If needed, consult an independent, qualified professional
Warning signs
- Very high valuations (forward P/E of the largest tech names well above the market average; Shiller CAPE near historic highs)
- High index concentration: a few stocks dominate "broad" world indices
- Growing gap between investments and the actual revenues of AI firms
- Increasing debt financing of data centres (sharply rising fund lending, according to the BIS)
- Large insider sales and sharp sentiment swings (e.g. the DeepSeek shock 2025)
Before · During · After
Before
- Check diversification: do not bet everything on a few names/regions
- Keep a liquidity buffer for several months
- Set the investment strategy in writing, matched to your time horizon
- Periodically bring positions back to their target weighting (rebalancing)
During
- Stay calm, do not abandon the strategy out of fear
- If possible keep up planned savings contributions (cost-averaging effect)
- Do not make decisions based on headlines alone
After
- Reassess risk profile and diversification
- Document the lessons: where was your cluster risk too high?
- Adjust strategy and rebalancing rules if needed
This really happened
-
2025 DeepSeek shock
On 27 January 2025, the low-cost AI model of the Chinese start-up DeepSeek triggered a sharp sell-off in AI stocks. According to Bloomberg/CNBC, Nvidia lost around 589 Mrd USD of market value in a single day – the largest one-day loss in US history. It shows how quickly high expectations can flip.
Source -
2000 Dot-com crash
After years of internet boom, the dot-com bubble burst from March 2000: the tech-heavy Nasdaq lost around 78 % of its value by 2002. Many highly valued companies of the time were barely profitable – a lesson in concentration and excessive expectations.
Source
Frequently asked questions
How do I protect my pension (3a, pension fund, ETF) from AI concentration?
Check diversification: don't put everything into a few securities or regions. Keep a liquidity buffer for several months, keep your investment strategy in writing and align it with your time horizon. Regularly rebalance positions to target allocation. This is pension education, not investment advice.
Why is market concentration a risk for "broad" global investments?
Many 3a, pension fund and ETF products with global exposure track indices like the MSCI World. Because around one quarter of this index is in the seven largest US tech stocks, such products are more concentrated in a few securities than you might think. A setback in these shares can therefore affect even supposedly broadly diversified investments disproportionately – a concentration risk.
Is an AI bubble proven – and what do regulators say?
No, a bubble is not proven but a possibility; a collapse cannot be predicted. However, several regulators – IMF, Bank of England, ESMA, BIS and the US Federal Reserve – warn of elevated valuations and market concentration. An example of the downside: on 27 January 2025 Nvidia lost around 589 billion US dollars in market value in one day after the DeepSeek shock.
Official sources
- Finance: the AI bubble in detail Analysis, web of companies and price charts
- IMF – Global Financial Stability Report Warning of stretched valuations
- Swiss National Bank (SNB) Holds US tech stocks directly