Emerging markets: the AI sell-off and the value opportunity

Artificial intelligence has split emerging markets into two camps, and the gap between them is where Rakesh Bordia sees his opportunity.

Bordia, a co-portfolio manager on Pzena’s emerging markets focused value strategy, describes a quarter in which almost 90 per cent of the asset class’s returns came from a single sector.

“Almost 90 per cent of emerging markets returns came from the technology sector,” he says, “and within that, three stocks, Taiwan Semi, Samsung and SK Hynix, drove almost three-quarters of those returns.” Growth crushed value, and the chipmakers riding the AI build-out did the crushing.

A market split in two

The winners were spectacular, and Bordia was a seller into the strength. Samsung rode a memory shortage that pushed DRAM prices up three to fourfold in a year. Taiwan Semiconductor compounded on AI-driven demand and improving margins. Both were among the portfolio’s biggest contributors, and both were trimmed.

The reasoning is a caution in its own right: the memory boom is a commodity cycle, not a permanent re-rating. “We believe DRAM continues to be a commodity business which will realign to its long-term returns as these companies add supply capacity and there is demand response from its customers,” Bordia says.

At the other end of the market sat the casualties. The strongest detractors were IT services companies, Cognizant and Globant among them. Both sold down hard on the fear that AI will hollow out an industry that has long billed for people’s time. Bordia’s verdict is blunt.

“There is deep concern in the market that AI will significantly impact IT services, and as a result the whole sector was very weak in the quarter. We believe that’s a severe overreaction.”

The scale of the repricing is striking. Pzena’s own work has Accenture, Cognizant and Globant trading at 8 times or less of its estimate of normal earnings, priced, in the firm’s words, as if in terminal decline. Globant’s forward multiple has fallen 85 per cent from its five-year median, Accenture’s 64 per cent.

Victims or enablers?

The bear case is not imaginary. Price deflation is real and accelerating. Contract renewals that once carried annual step-downs of 3 to 4 per cent now run closer to 6 to 7 per cent. That shift reflects a simple reality: the same job that once needed ten engineers now needs seven.

AI also threatens to disintermediate the industry. Work is moving in-house, software vendors are pushing downstream, and AI-native startups are undercutting the old advantage of employing thousands of low-cost engineers.

Bordia’s argument is that the fear has outrun the evidence, and that these firms are better placed as enablers than victims.

“Most companies need strong partners in implementing AI deployment,” he says, “and these IT services companies, which have strong domain expertise, should be strong partners in implementing AI solutions.”

He points to Globant as an example, “It has strong domain expertise, boasts a strong talent pool in IT services in Latin America, is in the same time zone as the US, has language capabilities beyond English, and so is very well positioned to help its customers make the journey of implementing AI solutions.”

The company detail supports the case. Cognizant draws roughly 60 per cent of revenue from healthcare and financial services. Its TriZetto platform processes about two-thirds of US healthcare claims, systems that no one can simply hand over to an AI tool. Its trailing bookings have risen 11 per cent on a year ago.

Accenture earns revenue per employee near $90,000, almost twice India’s largest outsourcers. It also returns about 12 per cent of its market value each year in dividends and buybacks. All three convert 90 per cent or more of earnings into cash. They carry conservative balance sheets, which is what buys an out-of-favour business time.

Value in dispersion

None of this makes the transition painless, and Bordia is not claiming every name wins. The most exposed firms, on Pzena’s reading, are those that simply rent out cheap engineering labour. The industry faces a difficult few years of pricing pressure before volume and new contract structures catch up.

The risk investors should weigh is that the cheapness persists. If AI disintermediation proves deeper and faster than the bulls expect, the cheap price becomes a value trap.

Fixed-price and outcome-based contracts are increasing to half of Cognizant’s revenue. These are the mechanism that lets a provider keep some of the productivity AI creates. They also hand the provider the risk of cost overruns.

What makes the setup attractive to a value investor is precisely the width of the valuation gap. “This momentum-driven concentrated market growth eventually gives way to broader participation,” Bordia says, “and this high dispersion is a perfect opportunity for us to identify great businesses at exceptional valuations.”

The same discipline drove two new positions in the quarter, both in unloved corners of China. KE Holdings, a property brokerage services company geared to a recovery in transaction volumes. Kunlun Energy, a gas distributor, bought cheaply into weak sentiment.

A timely reminder that concentrated, momentum-led markets create their own bargains on the other side. The chipmakers that carried emerging markets are riding a commodity cycle that will turn.

The market is pricing IT services names as though their transition will be hard, and paying investors to take that risk. It is giving them little credit, as Bordia sees it, for the chance these firms stay central to how enterprises adopt AI at all.