Emerging markets investors face a more complex AI opportunity

Emerging markets investors have spent much of the past decade being pulled between macro headlines, currency concerns, China risk and commodity cycles. In the first quarter of 2026, that familiar volatility was back.

According to Akhil Subramanian, portfolio manager at Pzena Investment Management, emerging markets rose through January and February before falling in March, primarily as a result of the conflict in Iran. But the more important may be what is happening underneath that index-level noise.

AI is no longer a single trade in emerging markets. It is creating clear beneficiaries in the semiconductor supply chain, while simultaneously pressuring software and IT services companies that investors fear may be disrupted by automation and AI agents. That makes the client conversation more nuanced than simply asking whether a portfolio has enough exposure to the AI boom.

AI is not one trade

Subramanian says it would be impossible to discuss emerging markets without talking about artificial intelligence.

“AI continues to be a theme that’s affecting EM primarily in the value chain of companies that are making and manufacturing semiconductor components, because that’s where you typically find EM companies.”

The way AI plays out in emerging markets is not the same story as in the US. In the US, investors may focus on hyperscalers, large software platforms and infrastructure spending. In emerging markets, the opportunity set is more likely to sit inside the hardware, memory and semiconductor manufacturing chain.

Samsung was a positive contributor for Pzena in the quarter, with Subramanian pointing to the ongoing strength of the memory cycle.

“There is very strong demand, and the three-player market of Samsung, SK Hynix, and Micron cannot meet it with the capacity that they have,” he says.

TSMC was also among the portfolio companies that benefited from semiconductor and memory value chains during the quarter. This alludes to a key nuance in EM portfolios, whilst they may carry meaningful AI exposure, it is often through different parts of the value chain. The risk is not only whether AI demand is real, but whether valuations have already moved too far.

The valuation discipline still matters

The harder part of the AI trade is knowing when a beneficiary has become too expensive. Subramanian says Pzena remains focused on “midcycle earnings power”, or the normalised earnings power of a business five years out, and then seeking to buy that business at a low multiple of those earnings.

That discipline can mean reducing exposure to companies that have performed well if the valuation no longer offers enough margin of safety. “When we see companies whose valuations have moved up and become more expensive on price to normalized earnings, they’re usually candidates to be reducing our position size,” he says.

This is especially relevant in the AI cycle, where share prices can move quickly and enthusiasm can compress future returns.

Advisers have seen this in developed markets, where the AI narrative has pushed a narrow group of companies to elevated valuations. Subramanian’s point is that the same valuation discipline applies in emerging markets, even where the underlying demand picture remains strong.

At the same time, AI is also creating dislocation away from the obvious winners. Subramanian notes there has been “a little bit of an air pocket around fears of AI disruption for the software industry,” alongside greater investor scrutiny of how much Google, AWS and Azure are spending on AI capital expenditure. “People are trying to understand whether some of these AI investments can earn an adequate ROI,” he says.

Disruption can create opportunity

That fear has opened a different kind of opportunity in businesses being punished by the market.

One example is Globant, an IT services company bought by Pzena in the first quarter. Subramanian says Globant sits “at the crosshairs of all the AI related disruption”, because investors are questioning whether AI agents will reduce the need for traditional IT services work.

Globant has historically been viewed as a premium IT services business, with expertise in user interface and user experience. Subramanian cites consumer-facing products such as the Disney app, Formula 1 and La Liga as examples of the kind of digital experiences developed and run by the company. But the sector is facing a demand downturn, and AI fears have added another layer of pressure.

“We believe some of it will be true,” Subramanian says of the AI disruption risk. But he argues Globant is also adapting, including through “AI pods” that combine agents and people to work on specific IT projects. For Pzena, the appeal is valuation as well as adaptation.

Subramanian says the company is trading at a very low multiple of midcycle earnings and, at the time of the update, on a double-digit free cash flow yield.

This is the second lesson for EM investors; AI disruption is not always a reason to avoid a company. It can also create an entry point if the market has overestimated the damage, underestimated the company’s ability to adapt, or pushed the valuation too low.

What advisers should take from the two-speed AI trade

The emerging-market AI opportunity is therefore becoming more complex. On one side are the semiconductor and memory beneficiaries, where demand remains strong, but valuation discipline is critical. On the other are software and IT services companies, where AI fears may be legitimate but potentially over-discounted.

Subramanian says Pzena’s process remains consistent during volatility: first, test whether any portfolio companies are structurally impaired, then search across a universe of around 1,500 companies for good businesses that may have been unfairly punished. The key question is whether the problem is temporary or permanent.

Across the wealth management landscape this might be a useful way to frame AI in emerging markets. It is not enough to own the theme.

The task is to separate capacity-constrained winners from overvalued beneficiaries, and genuine disruption victims from businesses temporarily marked down by fear. In EM portfolios, AI is no longer one story. It is a test of valuation, patience and stock selection.