Contrary to popular belief, what an active, fundamentally driven equity investor longs for most is not strong equity markets, but markets with breadth.
Is active management about to catch a “breadth”?
Contrary to popular belief, what an active, fundamentally driven equity investor longs for most is not strong equity markets, but markets with breadth. While no single definition exists for what constitutes a broad market, some point to the number of stocks that sit above their moving averages, others to what percentage of a market’s return or weight that is made up of the top five or ten stocks. Rest assured, an active manager knows one when they see it. Researchers at Goldman Sachs recently pointed out that market breadth far outranks economic growth or changes in interest rates as a predictor of mutual funds’ outperformance over their indices.
In the fourth quarter of 2025, markets finally, after many years of very narrow market leadership, started to display breadth. At the start of 2026, we have seen a return of dispersion. The magnitude of the difference between the winners and losers in the market came back with a vengeance – the re-emergence of both factors boded well for active management. With the onset of the crisis in the Strait of Hormuz, the market quickly reverted to very narrow leadership, where the AI trade was (yet again) the only game in town. While we in Q2 of 2026, were back to a narrow market, there are some tentative signs that market breadth might return in the medium term, and as we’ll touch upon, historical precedents that point to a longer cycle of market breadth that could mean easier times for stock pickers.
Market breadth in one visualisation
At least four interesting things seem to start be happening; geographical breadth is emerging for the first time in many years, the leadership within the hardware firms is starting to broaden out, usage of models is surging higher and in extension distribution has started to matter, and, finally, the political decisions regarding the war in the Middle East and AI regulation could help firms outside of the all AI theme start to perform.
1. Geographical breadth. Going back over the past fifteen years, the top five list of companies contributing the most to the return of the MSCI World Index have been dominated by US companies. Last year, Taiwanese TSMC made the list, but this was after a period of eight consecutive years where no non-US company was among the top five contributors. In total, only four non-US companies have made the list since 2012. So far in 2026, three of the companies on the top five list are non-US companies. While still early in the year, this points to a potential shift in geographic breadth. As US companies constitute about 70% of the market cap of the MSCI World Index, having managed a portfolio that is more prudently geographically diversified in the past few years, this concentration has been a headwind to returns – we could now potentially be in the early innings of a change in this dynamic.
2. Internal breadth within the AI hardware firms. As market leadership in the past few years has been driven by technology companies and most prominently among them the “Magnificent Seven”, Apple, Amazon, Alphabet, Microsoft, Meta, Tesla and Nvidia, this should be a good place to start looking in trying to understand this change in market leadership. Not only is market leadership changing, but technology leadership is also changing on the margin. Nvidia, despite the recent hype around new AI models and the surge in usage driven by coding and AI agents, has seen its shares trading essentially flat since early fall last year. Now other companies are starting to catch up with providing computing for the AI revolution. Amazon, for instance, recently reported that its own chip business now is a 20 USD billion business growing 100%+ per year. Compare this to 100 USD billion in compute sales; it might not seem significant, but Amazon’s chip sales are rising quickly. Compared to Nvidia’s peer AMD with 8 USD billion in CPU sales, suddenly a picture of new challengers emerges. While the absolute size still differs between the “old” semiconductor companies and the newly emerging competitors, this could lead to higher vertical integration for the cloud providers and fragment the profit pool available for the traditional semiconductor companies.
3. Distribution is starting to play a role. Markets are potentially starting to move away from only favouring the firms building the infrastructure of AI, to those also able to provide services and distribution. Historically, infrastructure layers in technology cycles tend to see rapid capacity buildout followed by pricing pressure, while application and distribution layers capture higher incremental margins due to customer proximity, switching costs, and differentiated user experience. We are now seeing Alphabet emerging as an AI leader. The company critically provides both the cloud infrastructure, as well as the services in the form of their own leading models with applications spanning from image generation, product development and orchestration levels for agents. The trend is starting to go global with Tencent’s shares rising 7% on the day they released their version of OpenClaw, proving easy access to orchestration for agents and mini-programs for their 1.3bn users. Since then, Chinese AI firms have fallen due to weak Chinese macro, but the signs are there that distribution will start to matter. While the top five contributors this year are hardware-focused, three of them are memory makers, this is pointing to the shift from training models to using them (inference), where memory is the limiting factor and in extension, where, in the next step, profits should start to accrue to the firms providing services on top of the infrastructure build out of the past few years.
4. Geopolitical and political tailwinds to market breadth. Finally, among the short to medium-term drivers of market breadth, we have to consider a reversal of one of the contributing factors behind the narrowness in markets starting in early April – a resolution of the crisis in the Strait of Hormuz. Normalising fuel and energy prices should help a struggling consumer and companies sitting outside of the AI theme. More recently, we have also seen political pressure on AI firms, both in the US with the administration shutting down access to Anthropic’s leading model and in China where the state seems to be taking a larger role in the roll-out of AI models. New AI regulation, while just emerging and at the moment seemingly arbitrary in the US, could, of course, slow down the AI firms reversing their leadership. The other side of this coin would be that firms perceived as disrupted by AI, most notably firms in the software sector, would see some pressure taken off and more time to adopt business models for an ultimately inevitable AI-driven technology cycle. The AI genie is out of the bottle, but geopolitics and domestic politics could alter the narrative’s trajectory in the medium term.
Zooming out from the quarterly and perhaps yearly market winners, another vantage point, the one of technological cycles, indicates that the market breadth could continue to expand in a longer-term perspective.
Carlota Perez
Finance Professor Carlota Perez, describes in her book “Technological Revolutions and Financial Capital”, how technology cycles reverberate through society in four distinct phases, divided into two periods: the “Installation Period” followed by the “Deployment Period”. The first phase, “Irruption”, sees funding for new clusters of innovation, where brand new industries are established as companies build new infrastructure. The next phase, “Frenzy”, sees increased speculation and financialization and asset bubbles start to inflate. What usually happens is that in the following “Synergy” phase, markets crash or correct, leading to political unrest, paving the way for regulation and wider societal adoption of the new technologies. Finally, as technology matures in the “Maturity” phase, it leads the way for a new surge in the next set of new technologies.
This pattern has repeated itself over many technological revolutions. The “Canal Mania” of 1793 saw widespread buildouts of canals in Britain. After the “Frenzy” phase and following crash, this led to the new set of leading firms not being the ones having built the infrastructure, the canals, but the multitude of industrial firms that could use these canals for transport. Almost the exact same pattern happened in the 1840s, as the railway mania led to the panic of 1847 and the diffusion of the infrastructure into broader society. The buildout
of the US highway network and the devastating crash in the 120 or so automobile manufacturers in the 1920s and 1930s led to suburbs, McDonalds, Walmart, AutoZone and thousands of successful businesses building on top of the previously established infrastructure. More recently, we remember the dominance of Nokia and Ericsson building the mobile infrastructure, never really recovering after the tech bust, only to see the next generation of firms dominating markets being Facebook’s, Amazon’s and Netflix’s building on top of the newly
established infrastructure.
While this time might feel different with the dominance of a few Cloud giants and AI-labs, history has proven time and time again that markets, after the initial narrow market focusing on the infrastructure build out broaden out. It’s always difficult to pinpoint exactly where we currently stand in relation to the Carlota Perez framework, but an educated guess, based on surging memory prices, upcoming IPOs of AI labs and the massive acceleration in the buildout of data centre infrastructure, is that we are in the “Frenzy” phase, with the
initial signs that the market is starting to broaden. Meanwhile there are tentative signs of short to medium term market breadth returning. Both are, of course, developments that we, as diversified fundamental stock pickers, welcome. Should we be proven wrong, with market narrowness continuing for some time and our exposure to the cloud providers now reaffirming their leadership, absolute returns might be strong, but alpha could be hard to come by. The crucial point is, though – historically it is not a matter of if, but of when, we get a prolonged broadening – and a return to a more robust environment for active management.
