Where could the next AI investment opportunities emerge?

By Lilian Chovin, Head of Asset Allocation, and Nathalie Malakee, Asset Allocation Analyst.

• Investing in artificial intelligence (AI) is no longer just about owning large US technology companies. Investors are increasingly distinguishing between those funding AI infrastructure, those supplying it, and those set to benefit from it.
• Equity markets are beginning to price a broader set of potential AI winners. Investors are looking across the value chain for companies that could benefit from the continued development and adoption of AI.
• Earnings visibility remains clearest among semiconductor and hardware companies, where strong AI investment is already translating into earnings growth. This supports our preference for emerging markets, particularly in Asia, where many semiconductor companies are benefiting from strong AI-related demand. 

The value of investments, and the income from them, can fall as well as rise and you may not get back what you put in. Past performance should not be taken as a guide to future performance. You should continue to hold cash for your short-term needs. This article should not be taken as advice. 

For the past three years, gaining exposure to AI was almost synonymous with owning US mega-cap technology stocks. That relationship is now less straightforward.

Equity markets have been anticipating a broadening of the AI opportunity into areas such as software, power and data centre infrastructure, yet earnings evidence continues to point to strength in the semiconductor sector.

Why the infrastructure buildout still dominates 

The scale of AI investment is increasingly visible in the accounts of the so-called hyperscalers, typically large technology companies that operate massive, globally distributed data-centre infrastructure. Analysts’ consensus estimates suggest that hyperscalers’ combined capital expenditure will exceed $1 trillion over the next 12 months. 

For hyperscalers, AI investment is recorded primarily as capital expenditure. For semiconductor manufacturers and associated industries, the same activity appears as orders and revenue. This distinction helps explain why the buildout of AI compute capacity continues to dominate the earnings landscape.  

Over the past 12 months, semiconductor earnings estimates have risen by around 90% relative to global equities. Almost 75% of US semiconductor stocks outperformed global equities over the same period, indicating that leadership has been broad across the sector rather than concentrated in a small number of mega-cap businesses. 

Unlike many thematic rallies, where valuations can run ahead of fundamentals, semiconductor performance has been underpinned by strong realised demand and rising earnings expectations. As a result, valuation multiples for many semiconductor companies have contracted as earnings have risen faster than share prices. 

Equity markets are looking to the next phase of AI 

Recent equity market performance suggests that investors are beginning to anticipate a broader set of AI beneficiaries, including software companies that could benefit from adopting the technology. After a difficult first quarter of 2026, US software stocks have rebounded strongly, with more than 75% of companies in the S&P 500 Software Industry Index outperforming global equities over the past three months. 

Some energy and real estate infrastructure companies also outperformed earlier this year as markets anticipated greater power, grid and data centre investment. 

The earnings evidence, however, remains more measured. Profit expectations for software, energy infrastructure and real estate infrastructure companies have been rising, but notably they have been increasing at a slower rate than the wider global equity market. This suggests that the anticipated AI benefits have not yet produced consistent earnings upgrades. While this does not necessarily mean the rotation is misplaced, a durable broadening would require earnings evidence, not price momentum alone. 

As evidenced in the table below, US semiconductor companies continue to show stronger price momentum and stronger profit growth than other AI-related US sectors.

Hyperscalers face a higher burden of proof 

The rotation within AI also reflects growing scrutiny of the hyperscalers and demonstrates the central tension within the current AI cycle. Hyperscalers are financing much of the infrastructure buildout but, unlike semiconductor suppliers, have not yet been its clearest earnings beneficiaries. 

While their profits are still expected to grow in absolute terms, earnings momentum has lagged global equities, capital expenditure has risen and expected free cash flow has deteriorated. This has led to a valuation de-rating of some companies, as investors demand clearer evidence that AI investment will generate attractive returns. 

Current earnings favour emerging markets 

The divergence in earnings across the AI ecosystem also has important implications for regional asset allocation. Hyperscaler exposure is concentrated in large US platforms, while the semiconductor supply chain is more global, with substantial representation across Asia and emerging markets (EM). A meaningful share of US AI investment is therefore recorded as revenue among companies in Taiwan, South Korea and other parts of the Asian technology supply chain. 

This reinforces our view to remain overweight EM relative to the US in our equity positioning. EM provides exposure to the part of the AI ecosystem where relative earnings momentum is strongest, alongside less demanding regional valuations. 

That said, the outlook is not without risks. A slowdown in AI investment could weigh on earnings momentum, while concentrated semiconductor exposure could amplify market volatility.  

We therefore carefully monitor the breadth of earnings upgrades across the sector, and whether AI spending is delivering tangible returns for the companies funding it.