Global stock markets are too high and set to fall, says Bank of England deputy

The key message is not simply that valuations are stretched; it is that the market is pricing a benign correlation regime that is most likely to break when shocks interact rather than arrive in isolation. That matters because the biggest air pocket is in crowded duration-sensitive growth exposures: once equity multiples compress, private credit marks, venture financing, and leveraged holders of AI infra all tighten together, turning a normal de-rating into a funding event. In that setup, the first-order loser is not just high-beta tech, but any business dependent on perpetual refinancing and low volatility assumptions.

The second-order effect to watch is supply chain leverage inside AI. If capex slows even modestly, the pain is concentrated in semis, networking, data-center power, and software vendors with usage-based monetization that depends on continued buildout; the downstream winners are more likely to be cash-generative software, cyber, and mature megacaps with genuine free cash flow rather than the highest multiple names. NVDA can still outperform in a shallow correction because it remains the cleanest earnings power in the ecosystem, but MSFT is more exposed to multiple compression as investors reassess the durability of its AI monetization premium.

The catalyst window is months, not days: a disorderly move likely needs a macro shock plus evidence of stress in private credit or liquidity-sensitive funds to force de-risking. Once that begins, vol tends to overshoot because dealer hedging and systematic de-grossing amplify downside, especially in names with heavy call ownership. The contrarian point is that the market may be underpricing the speed of transmission from private markets into public equities; the risk is less a slow grind lower than a gap-down repricing if confidence in shadow credit weakens suddenly.