Unlocking alpha: Navigating US equities

US economic growth continues to surprise on the upside, supported by resilient consumption and expansive fiscal policy, even as policy rates have remained relatively restrictive. However, this sanguine view masks more nuanced dynamics within the investment backdrop.

Divergence within the buoyant US equity market

The US equity market is buoyant. Aggregate corporate earnings growth remains robust and equity valuations have reached elevated levels, relative to both historical norms and our forward-looking measures of fair value. However, this aggregate view masks a nuanced story at a more granular level. Earnings growth has largely been concentrated among the hyperscalers - the large-scale multinational companies providing cloud computing services. Meanwhile, many stocks are increasingly being driven by optimism surrounding the outlook for artificial intelligence (AI), digital infrastructure, and automation.

Outside of these concentrated areas, a large proportion of companies have lagged the broader market, including many domestically-focused businesses. Combined, these dynamics have contributed to a significant increase in valuation dispersion across the market, as evident when viewed across the capitalisation spectrum (Figure 1).

Figure 1: Significant valuation dispersion across the US capitalisation spectrum

For illustrative purposes only. Past performance is not an indication of future results.
Source: Fidelity International, 31 October 2025. US large cap: S&P 500 Index, US mid cap: S&P Midcap 400 Index, Global ex-US: MSCI World ex-US Index. LC equilibrium FV: US large cap equilibrium fair value +/- one standard deviation.

Dislocation between equities and the macroeconomy

These trends underscore a profound transformation - investing in the US equity market is no longer equivalent to investing in the broad US economy, at least to the extent that it once was. Instead, the market’s return drivers have become more thematic, reflecting the increased influence of economic activity linked to innovations like AI, including investment and its ripple effects. Without the impact of this activity, overall GDP growth could be far lower.

The proliferation of capitalisation-weighted passive strategies of recent years has also driven a structural increase in return dispersion across the market (Figure 2), by indiscriminately funnelling a relatively high level of capital into overvalued stocks (and, vice-versa, a lower amount into cheaper stocks). In addition, the spikes around bouts of volatility and exuberance have become more aggressive, which creates risks but also increases the scope and magnitude of relative value opportunities for active investors.

A consequence of these combined trends is that the 10 largest US companies now comprise around 40% of the market by capitalisation and are responsible for half of its volatility. This is despite the fact that they only make up 35% of its aggregate net income (per the S&P 500 Index, two years to 30 September 2025) (Figure 3).

For illustrative purposes only. Past performance is not an indication of future results.
Source: Figure 2: Fidelity International, Refinitiv, 31 October 2025. The Cboe S&P 500 Dispersion Index measures the expected dispersion in the S&P 500 over the next 30 calendar days, as derived from the prices of options on the S&P 500 index and selected S&P 500 constituents using a modified VIX methodology. Figure 3: Fidelity International, data for two years to 30 September 2025. Data relates to the S&P 500 Index.

Software is eating the world

Addressing the investment challenges presented by this concentration requires a deeper understanding of what is driving it and whether it is justified. It is largely a function of the hyperscalers' success, which has generally been built on their ability to disrupt industries by delivering services more efficiently through software. In turn, this has endowed them with strong economic moats and huge growth runways, through which they have been able to deliver high returns on invested capital for extended periods, compounding to today’s huge size through reinvestment. This is the concept of software is eating the world. 

The optimism currently manifesting within markets is largely predicated on the fact that innovations will extend and accelerate this trend. Some savvy native cloud companies have already seen revenues at some of their key businesses accelerate (Figure 4), alongside a pick-up in worker productivity (Figure 5). Meanwhile, the threat of regulation also appears to have receded due to the deregulatory philosophy of the current US administration. From this perspective, high market concentration simply reflects reality.

For illustrative purposes only. Reference to specific securities should not be construed as a recommendation to buy or sell these securities and is included for the purposes of illustration only.
Source: Fidelity International, November 2025.

A new era of hyperscaler competition?

On the other hand, rapid technological evolution and AI model convergence could mean that maintaining economic moats will become more challenging and expensive for the hyperscalers in future. As an example, new AI models can be trained easily using those which already exist, as was the case with DeepSeek. As the capital costs associated with building them from scratch are much higher and switching costs are more limited, first mover advantage is also reduced.

The scale of AI investment being undertaken demonstrates this, while also suggesting a higher level of competition faced by the hyperscalers. This is perhaps unsurprising given that AI presents the most significant threats we have seen to some of their cash-cow businesses, in areas like search and web-based advertising. The technology arguably also holds the potential to challenge established competitive advantages like network effects and ecosystem leverage over the longer term. Creative disruption has always been a key characteristic of the technology sector.

If this is the new normal, then persistently higher capital expenditure requirements could act as an ongoing headwind against hyperscalers’ capital returns, a key pillar of their historical success. The effects would not be uniform on a company-by-company basis, but investment to date has clearly acted as a headwind against operating free cash flow growth for some of these businesses, even as their underlying cash generation has accelerated (Figure 6).

Figure 6: AI boosting hyperscalers’ revenues but higher investment a headwind

For illustrative purposes only. Reference to specific securities should not be construed as a recommendation to buy or sell these securities and is included for the purposes of illustration only.
Source: Fidelity International, November 2025. Note: operating free cash flow, which includes the effect of capex, has slowed as capex has risen.

The AI hype cycle

Seeing a technology which was only productised in 2023 driving improvements in real world enterprise workflows in less than two years later is staggering. Historically, it has typically taken at least 10 years before a new technology was battle tested and ready for prime-time enterprise.

Periodic surveys of our bottom-up analysts around the world show that companies have indeed already started to benefit from AI and are increasingly doing so over time. A survey in October 2025 revealed that nearly half of our analysts expect it to have a positive impact on the profitability of companies they cover in 2026 (Figure 7), up from up from just a quarter in 2025. 

Over the longer term, we expect AI to deliver significant benefits to corporates across sectors and industries. We tentatively quantified the potential effects in the ‘AI-Driven Renaissance’ scenario of our November 2025 CMA update paper, ‘dispersion in markets and the macroeconomy’. This found that AI could deliver a 3% annual boost to long-term US equity returns in an upside case. However, quantifying its ability to affect returns on investment remains difficult and we therefore plan to undertake further research into AI’s impacts as more evidence becomes available.

That said, the same October 2025 survey also showed that only a very small proportion of businesses are expected to derive significant profitability gains from AI in the near term (Figure 7). Even within the technology sector, where AI is already delivering significant benefits, the overall short-term effect on profitability is expected to be negative given the headwind of higher capital expenditures.

Market dynamics surrounding the AI theme should therefore be framed in context of the fact that technological revolutions often begin with exuberance, with corrections subsequently resetting expectations before the underlying innovation’s benefits are eventually delivered upon over the years that follow. Whether such a scenario plays out or not, evolving market dynamics associated with the theme will likely create mispricing opportunities for nimble active investors to exploit across different time horizons.

Figure 7: The financial benefits of AI are increasing but are likely to remain limited in the near-term

For illustrative purposes only.
Source: Fidelity International, Analyst Survey, October 2025. Question: “What impact, if any, do you expect AI will have on your companies’ profitability over the next 12 months?”

Is the hype warranted?

While the US equity market appears exuberant in areas, we are conscious that bubbles are only truly identifiable in hindsight when it transpires that the expectations that drove their creation were overestimated. The fact that superior benefits are currently being felt by savvy technology companies than their non-tech counterparts could simply be indicative of the future. If the benefits remain concentrated in this manner, the runway for disruption- and innovation-driven growth will indeed broaden and lengthen for businesses with the expertise to deliver AI services, which could justify some stocks’ current high valuations.

As such, it is possible that the accelerated speed of technology implementation means that this time is different from a market-innovation cycle perspective. It is prudent, however, to retain an element of caution - investors should remember that long-term success has more often been derived by buying assets cheaply rather than at expensive prices. 

For context, we have seen some instances of hyperscalers derisking their exposure to parts of the AI supply chain, against a backdrop where financing arrangements are becoming more circular and debt-fuelled, and physical and financial supply constraints being reached in areas. We also believe that some companies’ valuations are unrealistic in parts of the market. These include the space sector, the nuclear power complex, crypto treasury, quantum computing, debt-fuelled AI power, neo-cloud (including resurrected bankrupt bitcoin miners delivering GPU-as-a-Service), and other areas where unprofitable and even pre-product concept businesses are promising transformational change without verifiable roadmaps for how to deliver it.

Much has also been made of Nvidia’s unprecedented capitalisation. It is perhaps ironic that at US$5tn it has now surpassed that of the entire Japanese stock market, given that the Japanese Imperial Palace was said to be worth more than the entire state of California during Japan’s real estate bubble of the late 1980s. However, Nvidia’s exceptional earnings growth has kept its valuation multiple far lower than that of companies like Palantir and Tesla, which are trading on even higher PE multiples than Dotcom poster child Cisco did at its peak around the turn of the millennium (Figure 8). While the circumstances differ significantly in each case, it remains to be seen whether today’s more richly valued businesses will be able to deliver the earnings improvements necessary to allow them to grow into their valuations over time. Until they do, Dotcom analogies will be mooted.

Figure 8: Dotcom revisited? Cisco (lagged), Palantir (RHS), Nvidia, Tesla valuations

For illustrative purposes only. Reference to specific securities should not be construed as a recommendation to buy or sell these securities and is included for the purposes of illustration only. Past performance is not an indication of future results.
Source: Fidelity International, Refinitiv Workspace, 6 November 2025. Palantir, Nvidia, Tesla data from 06/11/2023; Cisco from 02/11/1998 - 2001 (Dotcom bubble peaked on 10 March 2000). Palantir’s recent sharp PE decline was partly a function of earnings improvement, demonstrating the potential for these companies to grow into their valuations over time.

Position thoughtfully, but don’t disengage

This doesn’t mean that US equities, which now form almost two thirds of the global equity market by capitalisation, should be ignored. To the contrary, there are many reasons to be optimistic about the market, especially on a selective, bottom-up basis where the macro backdrop is supportive.

That said, trade tensions are on an easing trajectory, fiscal tailwinds from the One Big Beautiful Bill are yet to be felt, and the Fed’s monetary policy stance is likely to become more dovish as it shifts its focus within its dual mandate. These trends are also occurring amid shifting dynamics in US dollar exchange rates, which have driven an improvement in the US’s terms of trade. This will boost reshoring opportunities at a time when government policy is aligning in support of US reindustrialisation. Combined, these factors could lead to a broadening of performance through 2026, as they arguably benefit domestically-focused Main Street businesses more than their multinational and thematically inclined peers, while many of the former are attractively valued relative to the broader market.

Even if a period of valuation contraction were to occur, this would not mean an absence of opportunity given the current level of valuation dispersion across the market. The history of the Dotcom era illustrates that strong innovation cycles can still play out during such periods, allowing leading firms to deliver impressive earnings growth. Indeed, our base case still projects a healthy average long-term earnings growth rate of around 6% for the US market and it remains possible that AI could drive significant profitability improvements for some companies more quickly than we currently anticipate.

Investors should remember, though, that risk management is as important to the generation of attractive returns as opportunity capture over the long term - especially through periods where valuations are high and return dispersion is elevated. A thoughtful, research-powered investment approach should therefore be employed to manage risk from the bottom up, rather than relying on a strategy that provides exposure to overpriced assets during periods of exuberance.

Market dynamics are creating a new age of alpha

The concentration of capitalisation, earnings and volatility makes idiosyncratic factors affecting the performance of the US mega caps more important, arguably enough to consider separately at the asset allocation level. Investors can therefore no longer assume that broad US market exposures are diversified sufficiently, especially given the correlation of thematic risk drivers among the hyperscalers.

Exposure to US equities still makes sense, but investors should be thinking about managing risk as much as capturing returns - through all levels of portfolio implementation. Given the difficulty in timing markets and potential performance-cost implications of traditional diversifiers, investments that can help manage risk from the bottom up should be prioritised. In other words, research-informed active management will be required as alpha becomes a more important component of risk-adjusted returns in the face of innovation, disruption, and high levels of concentration.

*Further information in German language.