Market drop-in October 2026
Headlines
The defining move of recent weeks has been in bond markets, and it is seen as a real yield shock, not an inflation shock. The US 10-year real yield peaked at 2.94% on 5 October, its highest since November 2008 and nearly a full point above where it started the year, with around half of that move since July alone. Of the roughly 110bps rise in the nominal 10-year this year, about 100bps is real yield as inflation expectations have barely moved. The market is repricing the cost of capital, not the price of goods.
Into that backdrop come three events. Chancellor Healey delivers his first Budget on 28 October with 10-year gilt yields around 5.4%, the highest since 2008, fiscal headroom squeezed to an estimated £12bn and around £10bn to find. Americans vote in the midterms in early November, where a divided Congress is the most likely outcome, and gridlock has historically been supportive for equities. And oil remains near five-month highs at around $108, underpinned by Hormuz disruption that cut Gulf crude flows to roughly 8.5 million barrels a day last week from over 20 million the week before.
Equity markets
Headline numbers remain strong, with world equities up around 13% year-to-date, but the story is in the past month and quarter. US small caps are down 4% on the month and around 7% on the quarter while large caps stayed positive, the first clue that this is a real yield story, since small caps are the most rate-sensitive part of equities. Japan is still the best performer year-to-date at around 21%, but the moves of 15 October wiped out almost a month of gains, and Asia ex-Japan fell 3.7% on the month, driven by the Korean and Taiwanese memory names. Tech has been the first casualty, and Asia is seen as the geared version of the same trade. The Nikkei and Kospi have become a leveraged play on the AI trade. We have been shaping our Asia exposure around this through the year. Gold and infrastructure, typically owned as hedges, have also fallen. These assets protect against inflation shocks.
Fixed income
Cash is up around 3% year-to-date and has beaten almost every bond line with a fraction of the volatility, while gilts are down 1.6%, so duration has continued to struggle. The flip side is that bonds pay again: investment grade yields are near 6%, BBBs around 6.2%, and high yield 8.1%. High yield spreads have widened around 80bps from July's low of 240bps. This is still below the 360bps seen in March when the Strait of Hormuz closed. Spreads are orderly, we see a repricing of the weakest borrowers, not a seizure of the market.
Global convertibles have sold off again. Their issuers skew to tech, so the past week's tech weakness has shown up inside these bond instruments. High yield now pays for two things, yield and a wider spread. High yield is also shorter duration than investment grade and so the asset class is less exposed to the real yield move, and its drawdowns have historically been around half those of equities. The repricing is concentrated in CCCs and below, where the AI-funded, refinancing-dependent borrowers sit.
A real yield shock, not an inflation shock
In 2022 every component of inflation was high, with core at 5.9% and food at 10.4%. Today headline CPI is 3.7% but core is just 2.6%, and the gap is almost entirely energy, up 20% over twelve months, with gasoline up 34% and fuel oil up 74%. Core goods are running at just 0.7% despite elevated oil, so there is no pass-through into goods prices. The simple way to explain this to clients is that prices are rising because the economy is running quickly, not because central banks have lost control of inflation. That is a good problem to have, but it still requires central banks to tighten, and it changes what works in portfolios as real yield shocks hurt long-duration assets and reward cash and short duration.
The wall of AI borrowing
AI-related debt supply has reached $489bn this year to mid-October, up roughly 50% on 2025, and hyperscalers have gone from under 1% of global investment grade issuance in 2024 to around 10% now. Most of this debt is long-dated, so it competes directly with government bonds for the same buyers. This is the mechanism linking AI capex to the gilt market, and it raises the cost of money for everyone, including governments. This is not a judgement on AI itself, as the biggest borrowers have very strong balance sheets, but much of the debt is secured against chips with far shorter lives than the bonds financing them, which is why credit, rather than equity valuations, has long been our bigger worry.
Power: The mechanism for a credit shock
US data centres need 97 gigawatts of new power by 2028, but only around 65 gigawatts has a credible path to delivery, a 33% shortfall that money alone cannot fix, since grid connections take four to seven years. The hyperscalers sign long-term power deals and build their own generation, but smaller neocloud firms borrow heavily to buy chips, land, and power that earn nothing until customers rent the capacity. CoreWeave is the case study: $50bn of debt against $5.1bn of revenue, with bonds now yielding around 16% and shares down more than 15% last week. Gartner expects 80% of neoclouds to cease independent operations by 2030.
If the power is late, the debt is still due. This is a credit problem, not an inflation problem. Even if AI demand proves as large as the optimists believe, the weaker borrowers can still fail because they cannot get the electricity to serve it. Credit crunches take on a life of their own, and clients tolerate falls in equities far better than sharp falls in assets they were told were safe. We saw this risk 15 to 18 months ago and built the bond book around it, introducing flexible managers who can avoid defaults in high yield and manage duration. We care about the quality of the borrower, not just the extra yield on offer.
Positioning
This leaves us cautious and more active in fixed income, with managers who can adapt to credit and duration risk, and thinking about equity exposure through a factor and thematic lens rather than a purely geographic one. Fundamentally, these are good problems, they reflect real economic momentum in the US economy and extraordinary productivity gains. The AI build-out is arguably the greatest bet capitalism has ever made, and our job is to embrace the opportunity while understanding the profound risks within it and safeguarding client assets through the process.
Sources: FactSet, Bloomberg, Morgan Stanley, IMF, BLS, Reuters, Gartner, data to 22/10/2026
