Coronation Global Emerging Markets

This commentary is for the retail class of the fund.

PERFORMANCE

The Fund returned -2.3% during the third quarter of 2026 (Q3), behind the -0.4% return of its benchmark, the MSCI Emerging Markets (Net) Total Return Index. The Fund has lagged the benchmark materially over one year, which has fed through into the longer-term numbers. We are very aware that relative returns have not met investor expectations and we are not complacent about that. We believe the Fund today offers exceptional value that we expect will be realised in the fullness of time.

It has been nearly two years since the US-based hyperscalers announced the step change in their capital plans, with combined annual capital expenditure heading towards US$1 trillion, manifesting itself through unprecedented demand for anything and everything that is required for artificial intelligence (AI) enabled data centres. Some emerging markets (EM), notably Taiwan and South Korea, have been primary beneficiaries of this, with most of the hardware required being manufactured in EMs, and this has been the main driver of the ~30% one-year return of the asset class.

ACTIVE MANAGEMENT IN NARROW MARKETS

For many active managers, however, this has been a very tough environment since the market return has been driven by a small number of stocks – all of them direct or indirect beneficiaries of AI-related spend. As Figure 1 below shows, the proportion of stocks in the Index that outperformed the Index has fallen to the lowest level in the past 18 years (which is when the Fund was launched).

Specifically, the entire Index return of 29.2% over one year has come from the ~40% of the market that is AI-related, with the rest of the market returning almost nothing (Figure 2). Just three companies provided 76% of the total market return; a handful of other AI-related hardware companies provided the rest.

The market has, during the course of this AI-driven cycle, become increasingly concentrated, with the top three stocks alone now close to 30% of the Index (Figure 3).

In order to outperform, active managers like us would have needed 30-40% of their portfolios in just a few names. As fiduciaries of retirement capital, we generally do not construct portfolios with such excessive concentration and potential for skewed outcomes in either direction. The downside of this is that when the market returns are very narrow, one can underperform materially in the short term.

The market has re-rated TSMC higher (yet it is still only on 20x next year’s earnings) since its near-monopoly position on the most advanced chips the world requires seems almost impenetrable, but for memory makers like Samsung and SK Hynix (on 4x earnings), the industry’s cyclical nature and these multiples mean the market is partly pricing in profits that will decline over time. Supply will undoubtedly catch up over time, as it always does. If it happens sooner rather than later, the outsized profits these companies are earning will normalise faster. As a counter, there are more signs of longer-term contracts being signed between the memory companies and their customers. As a result of this, and taking into account the significant declines in the share prices of Samsung and SK Hynix in July, we believe all three stocks now offer reasonably compelling forward-looking returns, with a combined weight for them in the Fund at quarter end of 18% (10% in TSMC and 8% in Samsung/SK Hynix combined). This is not an insignificant part of the Fund (although as shown above, the Index in just these 3 stocks is 30%), and in the case of TSMC, we have held a 9-10% position for some time. Mandates prevent the Fund from having more than 10% in any single stock. The more relevant valuation metric for the memory stocks for us is not the 4x earnings but the multiple on what we consider to be normalised earnings, and in this regard SK Hynix is now on 15x our estimate of normalised earnings and Samsung on 13x our estimate of normalised earnings. The Samsung preference shares (which is the class the Fund owns) trade at a discount to the ordinary shares, resulting in the preference shares trading on 10x normal earnings. All-in, we have around 40% upside to FV for SK Hynix and 50% for the Samsung preference shares, warranting the mid-sized position in memory in the Fund (see more detail in the next section).

Our view is therefore not always to ‘avoid AI-related stocks’ but rather that ‘there is far more compelling value elsewhere in the market from stocks that have been left behind’. Within AI, our view is that many of the next-tier (beyond TSMC) Taiwanese supply chain companies are overearning and expensive. At quarter end, the Fund had around 24% in AI-related stocks compared to their 40% weight in the benchmark; hence, we are still very underweight as we remain cautious on the AI capex cycle and its potential longevity, certainly from 2028 and, in particular, beyond this. The 24% is made up of 18% in TSMC and Samsung/SK Hynix (as referred to above), a 2.5% position in ASML, and the remainder is 3.5% exposure across three small positions in the Taiwanese AI supply chain stocks.

CONTRIBUTORS AND DETRACTORS

In the quarter under review, the memory stocks performed poorly. Having peaked in late June, SK Hynix fell by more than 40% during July before recovering somewhat to leave it down 23% for the quarter as a whole. Samsung declined by a third from its peak at one point and ended the period down close to double digits. As a result of our underweight positions in both these stocks, they were the two largest contributors to relative performance in the quarter, with a combined 1.9% impact. We had been reducing the positions in June as the share prices went vertical, in part driven by leveraged buying by South Korean retail investors. Before the sharp sell-off in July, the Fund was down to having a 3.5% combined position in the two memory stocks. The significant share price declines in July brought both into buying territory again, resulting in the current 8% memory exposure in the Fund.

Other notable positive contributors were Sea Limited (gaming, ecommerce, and financial services; +0.4% impact), Eternal in India (food delivery and quick commerce; +0.3% impact) and Brazilian oil producer PRIO (also +0.3% impact).

Four stocks detracted 0.5% or more from relative performance. The largest of these was Chinese battery maker CATL (-1.1% impact). Sentiment towards the stock has cooled – one reason being that two of the Chinese electric vehicle (EV) companies appointed an additional battery supplier. This supplier is very small and doesn’t have any impact of note on our earnings forecasts for CATL. Additionally, a number of the Chinese EV car companies have historically encountered difficulties when trying to diversify their supplier base (or move battery production in-house). Besides production difficulties, in the most recent case, the customer response was negative, resulting in a necessary change of this strategy. At a higher level, CATL is also actively exploring new technologies (such as sodium-ion and next-generation LFP batteries) to allow it to keep its cost advantage intact. CATL spends 3x more on R&D than the next biggest competitor, generates significant free cash flow (in contrast to its several main competitors, none of whom generate positive free cash flow), and has a large net cash balance (in contrast, all of its several competitors have net debt). CATL now trades on a very attractive 12x next year’s earnings and we added to the position during the quarter – making it a 4% position today.

Other material detractors were Coupang (-0.9% impact), Grab (-0.8% impact) and Naspers/Prosus (-0.5% impact), which underperformed Tencent in the quarter as the discount widened. We added to each of these in the quarter as our view on their long-term attractiveness is largely unchanged.

PORTFOLIO ACTIVITY

There were a few new buys in the quarter: Delta Electronics and Elite Material (both in Taiwan), HDFC Asset Management (India) and South32 (global). These collectively reached just over 3% of the Fund at quarter end. Delta Electronics and Elite Material are both beneficiaries of the AI buildout. Both were bought after sharp share price declines in the July sell-off. Delta is one of the higher-quality businesses in the Taiwanese AI supply chain in our view, and it is the leading supplier of power and cooling systems for AI servers, where content per rack is rising as data centres move to higher-voltage architectures, while Elite Material is the dominant supplier of high-end copper clad laminate, the critical material in the circuit boards used in AI accelerators, with over 40% share of that market and a proprietary resin formulation that competitors have struggled to replicate. HDFC Asset Management is India’s largest active equity manager, a highly cash-generative business earning returns on capital above 30% with a long runway of growth as Indian household savings continue to shift from bank deposits into mutual funds, and weakness in the Indian market during the quarter gave us an attractive entry point. South32 is a diversified miner exiting aluminium to focus primarily on copper, but also zinc-lead. As a result of the sale of the aluminium assets, as well as the ramping up of production from its copper and zinc-lead assets, the contribution of copper will increase from 33% today to over 50% over the next few years, and zinc-lead from a few percent to 20%. Copper, in particular, is one of the most attractive commodities long-term in our view, with additional demand drivers increasingly coming from AI and clean energy, and with an attractive supply/demand profile (a deficit, with very long lead times for new mines).

To fund the new buys, and the increased exposure to existing holdings, a few positions were sold. Among these were our remaining positions in Delivery Hero and Pepco Group, largely because of strong share price performance in recent quarters and limited potential for outsized returns from here. We also sold out of Tencent Music Entertainment. In China there is continued market disruption from AI-generated music, which the authorities have shown little appetite to crack down on. This is negatively impacting paid streaming services, and in particular the incentive for people to migrate from free options to paid tiers. With a wider range of outcomes, the risk-reward was no longer in favour of retaining the holding when much higher conviction options are available.

OUTLOOK

The team has been travelling extensively in recent months, with trips to Brazil, India, South Korea, Taiwan, and China. Several other research trips are planned in the weeks ahead. We are very excited about the prospects for the Fund. At the time of writing, the weighted average upside was 95%, with a five-year IRR of 24% p.a. This is indicative of exceptional value available to investors. If one excludes the holdings in TSMC and the memory stocks, the remainder of the Fund is even more attractively valued, with closer to 120% upside. Within this, several of the bigger positions have >100% upside to fair value and/or five-year IRRs in the high 20s or 30s. This includes Nu Holdings (Latam banking), MercadoLibre (Latam ecommerce and fintech), PRIO (Brazilian energy/oil), CATL (Chinese/global battery leader), Prosus (stake in Tencent), AIA (pan Asia insurance), Coupang (South Korean ecommerce), Sea Limited (Southeast Asia gaming, ecommerce, fintech), HDFC Bank (India banking), and Bank Central Asia (Indonesia banking), all of which are 3%+ positions.

Insights Disclaimer
Comprehensive Fund Factsheets