A correction in artificial intelligence-related valuations would not necessarily be confined to technology markets. It could spread to private debt, leveraged funds, creditors of hyperscalers and financial infrastructure that relies on a handful of IT providers. This is the chain of events described by Andrew Bailey, chair of the Financial Stability Board (FSB) and governor of the Bank of England, in a letter addressed to the G20 finance ministers and central bank governors. He is not predicting the bursting of a bubble. Rather, he is warning that several vulnerabilities could be triggered simultaneously by a single shock or a combination of shocks.
The starting point for his analysis is the Middle East. Andrew Bailey believes that the financial system has so far absorbed the supply shock caused by the conflict and the pressures on the global economy. In his view, the resilience observed demonstrates the value of the reforms adopted following previous crises. However, this initial test has not eliminated the vulnerabilities. The conflict has exacerbated energy-related inflationary pressures. Commodity prices have fallen from their initial peaks but remain sensitive to news from the region. Volatility has spread to several asset classes.
At the same time, market interest rates have risen since the start of the conflict, making borrowing more expensive. However, valuations of risky assets remain high. “We cannot therefore afford to be complacent,” warns Andrew Bailey.
Various potential sources
The FSB considers the markets to be vulnerable to a potentially disorderly correction that could spread across borders. It first identifies the vulnerabilities in sovereign debt: high issuance volumes, shorter maturities and increased use of leverage by certain market participants.
Private credit represents a second source of risk. The FSB does not merely question the level of debt among firms financed by these players. It highlights the opacity of this market, its links with other parts of the financial system, and the potential mismatch between the liquidity of the assets and that promised to investors. A fund may thus hold loans that are difficult to resell quickly whilst offering regular repayment opportunities. If redemptions accelerate, the manager must raise liquidity or sell assets on unfavourable terms. Relationships with banks, insurers and other funds can then transmit the strain beyond the initial vehicle.
Valuations linked to artificial intelligence are the third area of concern. The FSB describes them as strained, but its warning goes beyond the issue of technology share prices. It focuses on the financing that underpins their rise.
Leverage is becoming increasingly prevalent in the equity markets. Andrew Bailey highlights the growing use of leveraged ETFs, as well as correlated investment strategies that track the same trends. Retail investors are also part of this trend. At the same time, entities using leverage, particularly hedge funds, are playing an increasingly significant role in the equity markets. Some of these players are also exposed to sovereign debt. A fall in one market can therefore trigger sell-offs, margin calls or liquidity needs in another.
The dual impact of debt and concentration
“As we have seen on several occasions in the past, rising debt is a sign that a financial cycle is reaching maturity,” writes Andrew Bailey. Leverage amplifies gains when markets are rising, but it “can also intensify losses when sentiment turns”.
The crux of the concern lies in the combination of this leverage with the concentration of investment. “The problem is not simply that investors are borrowing more,” insists the FSB chair. Andrew Bailey highlights the interplay between debt, high valuations and market concentration, “particularly the rise in cross-investments between artificial intelligence firms and hyperscalers”. This situation could “exacerbate a future market correction”.
Hyperscalers provide the cloud, computing and storage capabilities required to develop AI models. They make substantial investments in chips, power networks and data centres. AI companies rely on this infrastructure, whilst their suppliers, investors and financial partners in turn rely on the expected growth in demand.
His letter highlights a second risk associated with AI. This no longer concerns the funding of its expansion, but rather the capabilities of the most advanced models. According to the FSB, these so-called ‘state-of-the-art’ models are demonstrating increasingly sophisticated autonomy, problem-solving abilities and offensive capabilities.
Ineffective national borders
The immediate threat to financial stability lies in cyber risk. AI could alter the speed, scale and cost-effectiveness of attacks. A faster or less costly attack could affect multiple firms, undermine market confidence and exploit the concentration of services amongst a few technology providers. National borders would offer little protection. A disruption can spread via common service providers, shared infrastructure and cross-border financial activities. Differences between jurisdictions in terms of law, cybersecurity and resilience are themselves becoming a source of vulnerability.
The FSB anticipates an environment characterised by a higher number of security breaches and an accelerated pace of IT system updates. This race against time may create its own operational risk if procedures for modification, testing and recovery do not adapt quickly enough.
Financial institutions, market infrastructures and their technology providers must therefore prepare for scenarios in which several businesses or shared services are disrupted simultaneously. Andrew Bailey emphasises the ability to restore critical systems and data from ‘bare metal’ – that is, by rebuilding the IT environment from its hardware components following a major incident.
The FSB Chair finally notes that many jurisdictions lack adequate protocols to regulate the development, provision and deployment of the most advanced AI models. He calls for making their safe and responsible dissemination an international priority.
The CSSF is already on alert
In Luxembourg, the warning aligns with several priorities already set out by the Financial Sector Supervisory Commission (CSSF). In its fund monitoring programme for 2026, the CSSF cites leverage, interconnections, private debt, liquidity risks, the valuation of illiquid assets and cyber threats.
The regulator plans to monitor alternative funds and Ucits funds with the highest levels of leverage. It also intends to expand its testing of margin and collateral calls relating to derivatives and repurchase agreements.
The CSSF will examine the credit risk management processes of open-ended fund managers invested in private assets, particularly where they have significant exposure to private debt. It is also maintaining its oversight of the valuation of alternative funds holding less liquid assets, including continuation funds.
The cyber security aspect forms part of the implementation of the European Dora Regulation. The CSSF incorporates its requirements into the supervision of asset managers and monitors major IT incidents and significant cyber threats reported to it.
