During April, the Strategy outperformed the MSCI Emerging Markets index by 0.6% as markets rebounded following the Iran conflict sell-off.
The greatest contribution to outperformance came from semiconductor stocks in Taiwan and Korea, but we also had a strong contribution from Chinese industrials and were helped by our underweight in Brazil and India. This was partly offset by other Latin America exposures such as Mexico, as well as Indonesia and South Africa which continued to be pressured by the situation in the Gulf.
During the month we continued reducing exposure to the UAE which has removed the remaining overweight to the country and left us underweight to the region. We have also continued taking profits from some of our semiconductor and precious metals positions, which have been major drivers of performance. This has allowed us to reallocate some more towards China and India, and to build up positions in some lower profile markets such as Kazakhstan.
Overall, the portfolio maintains a modestly positive skew to the semiconductor sector, but we aim to build up exposure to more neglected areas of the market. We are very aware that equities have now bifurcated clearly into AI stocks and ‘the rest’. These two markets have very different sensitivities and drivers and are now trading somewhat independently of each other – this needs to be recognised by investors. A number of the North of South team worked in equity markets the last time this happened, during what became known as the dot-com bubble.
Between January 1998 and March 2000, the Nasdaq 100 rose over 350%, led by a handful of internet stocks. It subsequently fell 80%, once the bubble burst. At the time of writing, the Korean KOSPI index has risen around 200% since the start of 2025, led by a handful of semiconductor stocks. This neither tells us that the KOSPI has further to go, nor that the index will collapse in ten months. In this case, as in 1998, extreme moves mean that the market had to rapidly adjust to a new and disruptive reality. We know that during the dot-com bubble, it adjusted upwards far too quickly (it took 15 years to regain its peak). Even without this hindsight, it was clear at the time that valuations were not remotely supported by earnings, with many leading stocks lacking profits or business models. Those companies that did have profits could trade at absurd multiples, with the Nasdaq P/E exceeding 100x at the peak.
The KOSPI on the other hand, has not even kept up with earnings upgrades – in fact, the Korean market is trading at a record low P/E of around 7x earnings despite the rally. These earnings are very real cashflow. They are also largely unsustainable once the supply and demand for memory chips comes back into balance and DRAM prices collapse. Whether memory makers, Samsung and Hynix, are cheap or overvalued now depends on one’s view of the duration of the memory super-cycle. We estimate that the stocks are priced for it continue into 2029, which seems consistent with the trajectory proposed by most analysts. One may have a more, or less, optimistic view than this, but it remains a sane debate around recognisable valuation metrics. This is very different to the dot-com era when new valuation metrics had to be invented to justify ever higher stock prices.
When we look to the US and the potential upcoming listings of newish trillion-dollar companies like OpenAI, SpaceX and Anthropic, we can see much stronger echoes of the dot-com era. Whether these really go ahead and how these companies deliver on their lofty promises will be the major determinant in comparing the equity market of the AI era to the dot-com boom.
