TL;DR

The International Monetary Fund says a fall in AI share prices could ripple through markets worldwide, The Guardian reported on Tuesday 13 October. In its Global Financial Stability Report (GFSR), the Fund calls risks “elevated”, with AI valuations and the debt behind them a principal reason, and says Britain and the euro area look “potentially more insulated” than Asia, though anyone holding US assets would feel it. The warning lands the same morning Reuters reported banking supervisors stepping up their own questions about AI.

Three risks, one theme

The Fund names three “salient risks”: an AI correction, higher government borrowing costs, and a large cyber-attack. Together, it says, they are “plausible near-term catalysts for a sharp tightening of financial conditions”. Athanasios Vamvakidis, a deputy director in the Fund’s monetary and capital department, said at the launch that the rest of the world’s exposure to US assets, and to AI in particular, “is the highest it has ever been”. That supports growth “through wealth effects today”, he said, but a correction would carry far beyond the US.

The arithmetic behind the concern, as the Guardian relays it, is concentration and borrowing. The seven largest US technology stocks now represent over 35% of America’s S&P 500, against 12% ten years ago, and their combined value has grown by more than half in two years. The Fund sees “growing evidence” that AI’s rollout may not need investment on the scale hyperscalers are committing, and that the economic gains may be smaller than hoped.

Leverage and circular deals

Three sources of borrowing worry the IMF: debt raised by the AI companies themselves, hedge funds’ positions in AI, and borrowing by retail investors to buy AI shares. In a downturn, it says, these would raise the odds of forced selling and volatile equity markets. It also flags “circular financing”, in which stronger AI firms are tied to weaker partners. The Guardian reports the Fund has counted $300bn (£227bn) of such deals linking hyperscalers to weaker AI companies over three years, topped up by $650bn of debt, in return for $1.4tn of future commitments.

Looking forward

The report sharpens a warning the Bank of England made a fortnight ago, when its Financial Policy Committee said AI-related borrowing was spreading exposure across capital markets, and it reverses the tone of the IMF’s January assessment that the AI boom was driving global growth. The cyber strand is not abstract either: Reuters reports that UK supervisors are nudging the biggest banks towards recovery systems kept apart from their main technology. For UK firms, “more insulated” is relative. Pension funds and retail portfolios hold the same US names, and UK data centre projects are financed on the assumption that hyperscaler demand continues. In our view, the Fund’s doubt about whether that spending is needed at all is the line worth rereading.