The funds enter the final quarter of the year at the lower end of their risk asset allocation with a more defensive tilt to the portfolio. Although still constructive on the transformational effect accelerated computing is likely to have on markets and economies over the next decade, now is perhaps a time for reflection. It has been a record-breaking rally off the Liberation Day lows of only six months ago during which the funds have materially outperformed. The merits of maintaining exposure at the lower end of our growth asset allocation range for now and reducing some exposure to the more speculative area of AI are too great to ignore despite our long-term optimism on earnings, margins, the global economy and AI in general.

Global P/E multiples are now back at their highest levels (ex-Covid) of all time. Although equity markets can continue their 30-year trend of re-rating, that is beyond the scope of this article and the current timeframe we are considering. Even allowing for the strong earnings growth to come, a period of multiple digestion would be sensible. The US market is trading at a record low FCF yield. There is some confusion around equity positioning that is dampening some positioning statistics but excluding that it is clear sentiment is now frothy and unfavourable if not quite at post-Trump election levels. The concentration of excessive speculation is clearly in one area namely AI. CTAs being 100% long, risk parity 100% long and momentum exposures at multi-year highs all point to the dangers of a Minsky moment increasing the need for pre-emptive de-risking.

The AI Capex Cycle

Our ability to navigate the volatility of this AI capex cycle has been key over the last three years to generating performance for the funds. The closed-loop ecosystem of OpenAI, Oracle, and Nvidia risks jeopardising the credibility of the whole buildout in the short-term. The faith one has to place in OpenAI right here right now to deliver hundreds of billions’ worth of combined revenue to Nvidia and Oracle over the next seven or eight years is not one we think we can share with our investors at this point. Also, the sense that this is accelerating the arms race for other hyperscalers is problematic. The current run rate of investment will see them plough ever-increasing amounts of their free cash flow into capex perhaps even having to resort to other forms of financing. The weight of this AI buildout may eventually begin to bear down even on the broadest of the MAG7 share price shoulders in the form of multiple de-rating. From hyper scaling to hyper spending.

Positioning and Risks

The risk profile of this capex cycle is therefore transforming from a public-market cash-fuelled one to perhaps a private-market debt-fuelled one. The implied capex, capital and power requirements for the latter half of this decade following recent news announcements seem in the short term unobtainable and ripe for disappointment. The speed and concentration of OpenAI announcements and the subsequent share price reactions are not coincidental but hard to repeat (Broadcom, Oracle, Nvidia, AMD, Intel and Samsung to name but a few). OpenAI, freed from its Microsoft relationship ahead of its latest round of funding, had the incentive to adopt this shock and awe approach. From here, at the very least the cadence of announcements should settle. If markets tend to peak on good news, what do they do after the greatest news flow cycle in the history of any one sector?

The disruptive impact OpenAI will have on incumbents is also beginning to come to fruition. Historically, we have always spoken about owning the winners of each cycle. What does one do however if the winner is not listed? In theory everything from Google’s search revenues to Amazon’s ad revenues is under threat in an agentic OpenAI world. We have concentrated our holdings in those on the receiving end of this capex namely industrials, power companies and semiconductor firms exposed to the infrastructure build and away from the hyper spenders.

Macro risks abound as they always do but a couple of issues seem particularly worrisome. They are well discussed but worth reiterating as any of them could trigger an unwinding of concentrated risk positions especially if sparked by a US dollar rally: French government collapse, UK budget risks, renewed Russian aggression or an extended US government shutdown. Ironically, the unprecedented strength of the above-mentioned AI capex spending may actually be causing a headache for the Fed. It is single-handedly propping up the US economy and driving recent GDP growth. This jeopardises the rate-cutting path the market has priced in. Overall economic growth and earnings especially in the US are solid and perhaps even accelerating.

Outlook

We enter the final quarter somewhat nervously near the lower end of growth assets. We acknowledge the next six months are seasonally some of the best for equity investors especially following a period of strength. At a sectoral level we are overweight industrials as they continue to benefit from the megatrends of digitalisation, decarbonisation and deglobalisation. The safety profile of utilities as well as their exposure to the power-hungry AI world makes them our second-largest overweight. We are also overweight in Staples and Pharma sensing some value in these much-beaten-up sectors though aware the market does not seem interested at the moment. We are materially underweight the MAG7 as we wait to see whether the market shares our concern about a switch from hyper scaling to hyper spending. We continue to be overweight the US dollar convinced the market is underestimating growth in the US and overestimating growth in Europe as well as overestimating political risk in the US while underestimating it in Europe.

The conditions remain in place for markets to continue their strong performance over the next five years as they have in the last. So too however do the conditions that have caused consistent air pockets of volatility.

Written by Philip Byrne, CIO, Cantor Fitzgerald Asset Management