Top
A narrow AI-led rally may dominate headlines, but we see a broader opportunity set across AI enablers, underappreciated beneficiaries and resilient franchises positioned for a range of outcomes.
In the second quarter of 2026, the MSCI World and S&P 500 indices staged their strongest quarterly advance since 2020, increasing 14% and 15%, respectively. Strong earnings growth (blended S&P 500 earnings per share (EPS) growing almost 30% year-on-year) and a partial unwind of geopolitical risks in the Middle East were the supporting factors behind the market strength. Yet the breadth of the market remains very narrow – fewer than 30% of S&P 500 constituents actually outperformed the S&P 500 Index during the second quarter.
Market returns continued to be supercharged by an exceptionally concentrated artificial intelligence (AI) trade, with semiconductor (up 55% in the quarter) and hardware (up 29%) companies contributing almost 60% of the MSCI World Index's total return during the quarter.
This narrow strength is making concentrated markets even more concentrated. Semiconductors and hardware now account for 30% of S&P 500 market cap, up from around 10% pre-pandemic and above the dotcom peak of 24%.
Capital flows have become similarly concentrated, with increasing retail participation and the growth of leveraged investment products directing an ever-larger share of each incremental investment dollar towards the same small group of AI infrastructure stocks, reinforcing momentum in an already crowded trade.
This concentration is not without earnings support, but the question for any long-term investor is whether those earnings reflect a durable step-change in industry economics or a temporarily favourable point in the cycle. Take memory, for example: Bloomberg’s Memory Exposed basket rose more than 130% during the second quarter as DRAM1 and NAND2 prices continued to surge. After 18 months of rising prices and constrained supply, operating margins across parts of the industry have moved from what has historically been a 0%-30% through-cycle range to 70%-80% − levels that exceed those typically achieved by even the highest quality software businesses.3 Economics this attractive inevitably encourage new investment and, in a largely commoditised industry, periods of exceptional profitability have historically proved self-correcting as additional supply (in this case potentially from China) catches up with demand.
The Quality AI opportunity
We firmly believe in the transformative effect AI will likely have on certain parts of the economy. We also think there are attractive investment opportunities in stocks that are helping to create the compute infrastructure needed for broad advanced AI adoption.
However, our focus on owning businesses capable of predictable earnings compounding through the cycle has precluded us from owning stocks where we think current earnings power looks unusually frothy and/or the underlying business is of insufficient quality. This has been challenging from a relative performance perspective, particularly as the extraordinary concentration of recent market returns has meant keeping pace with the index would require materially greater exposure to the semiconductors complex than we believe is consistent with our active investment philosophy.
Instead, we participate by owning select AI enablers and beneficiaries that exhibit the quality characteristics we have always sought: durable competitive advantages, pricing power, resilient revenues and disciplined capital allocation. For example, we own TSMC, whose leading-edge manufacturing capability represents a bottleneck in the semiconductor supply chain, benefiting regardless of which chip designers ultimately capture the greatest share of AI compute; Synopsys, whose mission-critical software tools benefit from resilient semiconductor research and development budgets and increasing chip complexity; and Schneider Electric, where the evolving power architecture required to support the rapid expansion of AI data centres is creating long-duration growth opportunities. We also feel favourably towards the hyperscalers, Microsoft, Amazon and Alphabet, which are the critical providers of compute needed to run AI applications. While the scale of current AI investment raises questions around the risk of excess capacity, we believe their diversified cash-generative franchises, strong balance sheets and multiple avenues for monetisation should position them to earn attractive long-term returns. Importantly, each remains an exceptional business even absent the AI opportunity.
Opportunities beyond the current hype
While the rising AI tide has propelled a relatively narrow, if growing, area of the market, it has left many other quality businesses behind. Since mid-2025, parts of software, professional services and non-balance-sheet financials have experienced substantial deratings as investors reassessed companies' long-term competitive positioning and terminal values in an AI-enabled world – despite, in many cases, continued robust operational performance and healthy earnings growth. Our exposure to these perceived "AI victims" has been a significant driver of both absolute and relative underperformance over the past 12 months.
Over this period, we have substantially refined our assessment of AI risk across the portfolio, scrutinising the durability of competitive advantages and long-term earnings visibility across different business models. We have focused on four broad areas of risk: workflow, data, consumer and token. Of these, we have found workflow-orientated software to be the area where AI has most meaningfully increased uncertainty by lowering barriers to entry and widening the range of competitive outcomes. As a result, we have materially reduced or exited positions where our conviction in long-term earnings compounding has diminished, while retaining exposure to businesses we believe possess more resilient competitive advantages, including systems of record and enterprise platforms. We also remain constructive on businesses built around proprietary or contributory datasets, consumer ecosystems and transaction infrastructure, where competitive advantages appear more durable and, in some cases, may even be strengthened by AI.
Importantly, we do not believe every business that has derated because of AI fears is equally vulnerable to AI disruption. Across our portfolios, businesses have continued to compound revenues and earnings despite sharp multiple compression. Many companies continue to offer the prospect of mid-teens shareholder returns, supported by high-single-digit revenue growth, modest margin expansion – potentially enhanced by AI-driven productivity gains – and attractive free cash flow yields. Unlike today's AI infrastructure trade, where elevated earnings are increasingly accompanied by demanding valuations, a number of these businesses now trade at prices implying materially lower long-term earnings power than we believe is justified. We believe rerating is likely to come as companies demonstrate that AI is not only reducing costs, but also creating incremental revenue opportunities through enhanced products, deeper customer engagement and improved monetisation.
Opportunities outside of the AI debate
We are also excited about opportunities outside of the AI theme. In recent quarters, we initiated positions in luxury franchises Hermès and Ferrari, where short-term cyclical concerns created attractive entry points into businesses with exceptional pricing power, scarcity-driven demand and some of the strongest competitive advantages in global consumer markets. We also added McKesson and Cencora, whose mission-critical roles within pharmaceutical distribution underpin resilient earnings growth, strong cash generation and attractive returns on capital.
Attractive opportunity in absolute and relative terms
We consider our portfolios to be deliberately positioned for today’s environment. They combine selective exposure to AI through carefully chosen enablers where we believe competitive advantages are durable, with a collection of high quality businesses whose long-term fundamentals remain intact despite substantial AI-driven deratings. As these companies increasingly demonstrate that AI can enhance, rather than impair, their long-term earnings power, we believe there is significant scope for valuations to recover. The portfolios also maintain a meaningful allocation to defensive, high quality businesses – including long-standing core holdings and newer opportunities in areas such as luxury and media – whose investment cases are largely independent of the AI debate, and which should provide resilient earnings through the cycle.
Relative to the broader market, we believe the portfolios offer more credible earnings growth, underpinned by superior top-line growth rather than relying on further margin expansion from already elevated levels. They continue to exhibit the quality characteristics we have always sought – including higher returns on operating capital, stronger gross margins and lower leverage – yet today trade at an unusual discount to the market in free cash flow terms.
Importantly, the portfolios are deliberately positioned across a range of potential outcomes. If enthusiasm around AI infrastructure persists, we continue to participate through our considered exposure to hyperscalers, semiconductor bottlenecks and other high quality enablers. If sentiment broadens, we believe many of our oversold holdings have significant scope to rerate as they demonstrate AI-enabled revenue opportunities alongside the productivity benefits already emerging. And, if the market's focus shifts away from cyclical growth altogether, we would expect the portfolios’ resilient earnings profile to once again prove their worth.
Source for data cited, unless otherwise stated: MSIM, FactSet, as of June 30, 2026.
1 Dynamic random-access memory
2 NAND flash memory is a type of non-volatile storage technology that can retain data without a power source.
Featured Insights
