View From The Top: When good is no longer enough

View from the very top: Market rotation and investor scepticism are both natural and healthy. However, there is no getting away from the fact that the artificial intelligence boom has been holding up both equities and the economy. A slower pay-off from AI is a problem for everyone given market concentration levels and heightened liquidity pressures. We may now be moving into a more challenging phase for the AI trade. If the AI cycle does not collapse but only ceases to surprise, then we may need new things to take the market higher. At least the economy remains robust for now, even if the next move for US interest rates will likely be upwards. Higher bond yields would be a problem for equities, especially given their high starting point after a recent period of very strong returns. We may still be waiting some time for AI to deliver deflationary productivity gains. For now, continue to diversify proactively and remain opportunistic.

Asset Allocation:

Equities: July was characterised by a massive and brutal reversal in the momentum trade, with the monthly unwind between the market’s winners and losers being the largest since the COVID-19 pandemic (per Bloomberg). Diversification is trumping concentration and value is outperforming growth. This is best evidenced by the performance of equally weighted versus market capitalisation-weighted indices. We have long argued that for the equity market rally to be sustained, it needs to broaden. Rotation is entirely normal and we favour active approaches to equity selection. Small caps and healthcare look interesting. Opportunities will also emerge from the recent sell-off in technology. Earnings upgrades will also continue to matter, even if expectations are now higher than at the start of the year.

Fixed income/ Credit: Government bond yields continue to move upwards across the world in anticipation of potentially tighter monetary policy. 10-year US Treasury debt yields 4.7% versus less than 4.0% in late June. 30-year yields are the highest since 2007. Investors are requiring higher returns in a more uncertain world. At the same time, there is massive corporate credit issuance underway to fund the artificial intelligence boom. Investment grade and high yield spreads remain tight for now. However, higher yields will present a pain point for borrowers attempting to refinance maturing obligations undertaken in an era of easier money. Our allocations within the space remain highly selective.

Currencies: We continue to see the path of least resistance for the US Dollar as higher. Superior US economic and productivity growth relative to the rest of the world would argue for a stronger currency. Even if Dollar strength may be secular, this does not rule out periods of mean reversion within a broader upward trend.

Gold: After the worst quarterly performance for gold in a decade during the three months to the end of June, we believe that there is a case for being more constructive at current levels. While higher Treasury yields and a stronger US Dollar could continue to pressure gold in the near-term, we believe the precious metal remains a crucial long-term core asset and portfolio diversifier.

Alternative assets: Hard assets perform an equally important role in portfolio diversification, in our view. Not only do they offer income-generating characteristics, but they provide resilience in the face of uncertainty. It is important, however, to remain selective across this asset class, favouring quality. Our focus is centred particularly on uncorrelated strategies.

High starting points are always a problem. With the MSCI World Index having delivered almost 12% annualised returns in the last five years and over 21% since its October 2022 trough, a broad-based market rally looks less likely at this stage of the cycle. Elevated earnings expectations and valuations do not help (not to mention excess leverage in certain segments of the market). Timing market tops is never easy, but now is the time to start rethinking approaches to diversification.

Continued earnings momentum has supported equity markets since the start of the year. Although S&P 500 Index earnings are forecast to have grown at over 20% year-on-year this quarter – the fastest pace in over four years, according to Bloomberg – it is worth bearing in mind that the Index has been flat since mid-May. Recent market moves, however, hide some important dynamics at work below the surface. An equally weighted version of the S&P 500 Index currently sits at an all-time high. Tech proxies (the Nasdaq and Philadelphia Semiconductor Indices), by contrast, have witnessed significant recent meaningful drawdowns. Momentum is yielding to gravity. Derisking, and banking profits in prior winners, seems only logical. Now is not a time to panic.

It is very rare for market cycles to peak when both earnings and employment trends are positive. When profits are high and rising, there is a logic to lean-in. 90% of S&P 500 Index companies have seen earnings upgrades in the last three months, per Bloomberg. Any market rotation also clearly creates opportunities. The latest Bank of America Fund Manager survey shows that investors are net long equities by the largest margin since December 2024. Consider also that some $7tr of cash remains in money market funds (according to Bloomberg), waiting to be deployed.

Markets have shown a continued ability to climb metaphorical walls of worry, be they relating to the economy and Central Bank policy, or continued geopolitical uncertainty. Perceived robustness, of course, can easily morph into complacency. Have no doubt, the artificial intelligence boom more than any other factor has been sustaining both the economy and the stock market.

While AI enthusiasm has so far supported equities and justified ever-higher infrastructure investment, rising credit risks, intensifying Chinese competition (first DeepSeek and now Kimi K3), plus the growing scrutiny of AI monetisation could mark the beginning of a more challenging phase for the AI trade.

A slower AI pay-off (or time to generate a return on invested capital) is a problem for everyone for two reasons. First, an enduring mega-cap tech sell-off would most likely initially take the rest of the market with it. Next, stretched balance sheets and growing credit risks could result in tighter financial conditions and pressured liquidity across the broader economy. For all AI’s promise, the path to realising significant productivity improvements is far from straightforward

The sustainability of extra borrowing costs combined with growing capital expenditure requirements has emerged as the largest near-term concern. Put simply, consensus estimates for hyperscaler capex are going up and free cashflow estimates going down, as Bloomberg data show. The five major hyperscalers are forecast to spend at least $1.1tr on AI infrastructure build per annum over each of the next three years, according to Morgan Stanley. For context, the US Department of Defence has proposed a budget of $961bn for fiscal 2026.

Against this background, bond issuance by major US tech companies has exploded, from just $20bn in 2024 to $120bn last year. In the first half of 2026, the five largest hyperscalers raised $160bn of new credit, per Bloomberg. Investors should take note when previously asset-light cash-generators are now becoming major investors in artificial intelligence infrastructure. If these businesses (and other related players such as Anthropic, OpenAI and SpaceX) cannot earn sufficient returns on AI spending, then funding may dry up rapidly, amplified by interconnected vendor financing arrangements.

Optimists will continue to assert that demand for new compute is effectively insatiable, driven by the growth of inference, the rise of agents and ever more powerful AI models. However, investors should not dismiss an alternative outcome. Better efficiency, model commoditisation (as Chinese LLMs gain in credibility) and slower enterprise adoption could constrain demand at a level well below current expectations. The result may be an industry that has built capacity faster than demand develops, leaving returns on infrastructure investments under pressure.

At the very least, the market has become highly sensitive to any perceived change in the AI narrative. Some scepticism is, of course, healthy. The time to start worrying is when no-one questions the narrative. For us, a plausible middle ground is that the AI cycle does not collapse, but that it ceases to surprise. Businesses will continue to assert that demand remains strong, simply because they must. A more nuanced reality is that the market may be beginning to correctly reprice hyperscalers more as utilities.

In this world view, we may need new things to take the market higher, since markets typically need new narratives. At least the economy remains broadly robust. The current US economic cycle is currently the sixth longest since 1850 and would become the fifth longest if a recession is avoided in 2026, according to data from Deutsche Bank. Manufacturing output and retail spending are both expanding while last month’s input cost and inflation prints were below expectations. The IMF assumes 2.3% GDP growth for the US and 3.0% for the world in 2026.

Two factors could undermine the economic outlook. First, there remains uncertainty over the direction of Fed policy. Even if inflation has fallen from its peak, a 3.5% CPI figure remains well above the US Central Bank’s target. Rhetoric and a commitment to “price stability” (the mantra of Kevin Warsh) may not be enough to assuage markets. Three of the twelve voting members of the Federal Open Market Committee supported a rise at July’s meeting, when rates were kept on hold. Fed Fund Futures now discount around a 70% chance of an interest rate hike in September – there is no meeting in August – with further hikes to come in 2026 and 2027. Tighter financial conditions would impact both consumers (the personal savings rate is the lowest since 2008, per data from the Bureau of Economic Analysis) and businesses.

A febrile Middle East may also influence not only the Fed’s thinking but economic confidence, even if the world has clearly proven itself to be more resilient to oil shocks than previously thought. The US remains unable to extract itself from a war it helped to start five months ago. Recent events point to a material deterioration relative to June’s peace agreement, with traffic in the Strait of Hormuz at a virtual stand-still and conflict now spreading to the Red Sea. The scope for misunderstanding remains significant even if rising military, economic and political costs ultimately favour compromise. Geopolitics may remain a manageable risk, but the way forward can still be fraught.

An improving macro path is needed to boost the fundamental floor higher. Inflation, arguably, holds the key. If it (re)accelerates, then bond yields will likely rise and equities fall. The inverse is equally true. Maybe, it’s enough for investors if growth remains resilient and inflation expectations fall, with there being scope for monetary policy to ease at some (distant) future stage. Should AI deliver the much-vaunted productivity gains that the technology promises, then the economy would be the ultimate beneficiary. Time will only tell. For now, the best strategy remains one of proactive diversification.

Alexander Gunz, Fund Manager