Emerging markets: on the path to the industrial revolution 5.0

Rob Mumford from GAM Investments’ Emerging Markets Equity team discusses the recent surge in technology stocks, discussions around generative artificial intelligence (AI) and implications for the emerging market technology sector

27 July 2023

• The digital progression from information to actual intelligence is driven by a combination of software and hardware advances, with emerging market (EM) Asia playing a key role
• Breakthroughs in generative AI are positive milestones, and only one leg of digital 4.0 on the path to the industrial revolution 5.0
• Not only is the technology (software, hardware, firmware) a secular growth story but the applications and services that feed off this stack. These include increasingly home-grown software in Asia ranging from the Indian consultants to the China software providers across a wide range of services (ERP to cyber security)
• With the EM technology sector at 21% of the MSCI Emerging Market Index and growing fast in a number of segments, we believe emerging market equities are a great way to play this digital evolution

Excitement over generative AI is warranted, though it is just one aspect of ‘Digital 4.0’ (as coined by the GAM Disruptive Growth team). Digital 4.0 is the next phase of digitisation powered by new applications incorporating cloud, AI, edge technology and a maturing metaverse with many years to run. The issue is that new digitised products and services are generating data at an exponentially increasing rate, feeding off and requiring increasingly strong compute power to run, interpret and actualise.

EM technology is the second biggest sector in the MSCI Emerging Market Index (at 21.2%, just a touch below financials at 21.9%) and we expect new and existing constituents to play a key role in Digital 4.0 given EM’s dominant role in a number of areas.

In a similar fashion to the US equity market, the EM information technology (MSCI) sector has significantly outperformed the index. It is up +19% year-to-date versus the index which is up +3% (as of 2 July 2023), though this is not quite as extreme as the US tech sector (S&P Information Technology) which is up +42% versus the index +16% (a large part due to Nvidia’s move up +189% year-to-date, while AMD is also up +76%).

Chart 1: The fifth industrial revolution involves the combination of humans and machines at work – the five waves of industrial revolution

For example, TSMC has around a 60% share of the foundry market which, coupled with Samsung Electronics’ share (of about 7%), takes Asian dominance in the segment over two thirds. In simple terms, a foundry company takes semiconductor designs from the likes of Nvidia and actually fabricates the design on a wafer to make the integrated circuit or semiconductor – so crucial to the advances and current capabilities.

There is a positive top-down impact from technological advances; PwC estimates AI alone will lift GDP forecasts by 26.1% in China and 14.5% in North America by 2030, with these two countries accounting for roughly 70% of the global impact. North America, with advanced technology, extensive datasets and fast adoption can support an above average positive impact, in our view. For EM, in China any limitations on advanced chip availability are offset by the large data set, considerable talent pool, large application opportunities (as a highly digitised economy) and policy driven adoption, driving the potential for above average GDP benefit.

Nvidia’s upwards revised guidance (on 25 May) was a major wake up call for the global technology sector. It showed how secular advances in technology are having a very significant impact, not just from next generation AI applications (including generative AI) but a number of other new areas (including electric and automated vehicles) with a positive impact on the associated supply chain. This is particularly relevant for large parts of the EM technology sector.

Chat-GPT alone requires just under 300,000 processors (280,000 CPUs and 10,000 GPUs) and Gartner expects the global AI semiconductor market to grow at 21% compound annual growth rate (CAGR) from 2023 to 2027. These processors are not acting in isolation; a key point to remember is that there is a huge requirement for a technology stack to support not just this application but the advanced technology and services of Digital 4.0 and beyond.

Electric vehicles and autonomous driving with AI as an integral give another glimpse into the expected massive future demand for chips, data processing and infrastructure that is required for the devices and services of Digital 4.0. A modern electric car has as many as 3,000 semiconductors, almost two times traditional cars. For autonomous driving, AI is crucial to allow self-driving cars to sense, perceive, navigate and make real time decisions based on data gathered from sensors. In a report from Bank of America, it quotes an Intel report estimating a connected autonomous vehicle will generate the same amount of data as 3,000 internet users, while two cars in communication (between themselves) will generate the equivalent of 8-9,000 internet users.

The growth and increasing range of goods and services has positive implications for companies across the full technology spectrum including semiconductors, semi equipment, memory (advanced server specific memory), substrate, motherboard packaging, and assembly. EM and Asian companies have significant exposure in these areas and have a dominant role in a number of segments.

Advanced memory chips (memory chips that store and/or process data) are absolutely crucial to AI given the need for faster computational capability (with less power consumption) versus pure storage capabilities. The memory sector has consolidated (after the merger of Elpida and Micron in 2013) into a three-player market, with two of the three Korean companies (SK Hynix and Samsung Electronics) having about a 60% market share. Memory chips are transitioning beyond DDR4 to higher performance chips including DDR5 and HBM which are critical to AI servers. SK Hynix has a technological lead in HBM chips which is expected to last for at least a few years. Samsung Electronics’ recent Q2 earnings also gave us an insight into the current boost from AI-related demand with its lowest result since Q1 2009, reflecting the current cyclical challenges. However, it beat forecasts due to demand for its high-end AI related products (DDR5 & HBM) and indicating the worst of the downturn may be behind us.

AI servers are rapidly taking off within the Asian original design manufacturer (ODM) supply chain who, due to a long history of R&D and client interaction, have circa 90% market share in this area. While the advance in processors and associated technology has led, or rather forced, foundry into advanced packaging, the market share in broader packaging and testing is still dominated by Asian players with just over a 50% market share. This was clearly in evidence during the Covid-19 lockdowns and one of the reasons for the significant chip shortage at that time.

For most of the related supply chain, AI advances are driving not just a positive volume story but also selling price (ASP) and margin story. The two segments above (memory and servers) are a good example to illustrate this. In memory, 1Gb of DDR4 is USD 0.2 while for HBM it is USD 1.2-1.3 and for 128GB DDR5 it is USD 0.9-1.0 – a 5-7x higher price in the fastest growing product. Similarly for the ODMs, an AI server is two to three times the price of general purpose servers and again the fastest growing segment. Given this double benefit, analysts (including Bank of America) forecast AI servers will account for circa 20-30% of total ODM server revenues by 2025.

In the MSCI EM Index, the key current weightings within the EM information technology sector are semiconductors and semi equipment at circa 10% of the index; technology hardware and equipment are just below this at circa 9% while software is a mere 2.3%. The software weighting is a massive opportunity as home growth technology and applications become more prevalent and is likely to provide a strong underpinning for the technology sector and broader index.

Chart 2: MSCI Emerging Market sector earnings growth (CAGR forecasts for ’24-’25)

Outside of the Indian consultants, which in this software segment of the MSCI EM Index accounts for circa 1.9% of the 2.3% weighting in software, China software and services has currently just a 0.3% weight in the index; China has a number of interesting opportunities across verticals including home growth office, enterprise resource planning (ERP) and cyber security. China currently forms a small portion of the sector but the national promotion of, and drive to, increase self-reliance in key technologies will likely see strong growth in all segments.

S&P/S&P Tech and EM/EM Technology forecast earnings (CAGR) growth over ’24-‘25

We have been favouring information technology in anticipation of a broader technology hardware cycle bottoming over Q2 while remaining positive on structural themes relating to Digital 4.0.

We do think investors should be cautious given some strong moves to date, current valuations, the cyclical backdrop of technology (which remains difficult) and also the fact some AI-related products may only account for a small percentage of companies’ current business products (though in general these products are growing the fastest).

For example, TSMC is expecting Nvidia’s success to lead it to grow from a 6-7% to a 10% client according to Credit Suisse analysts. It expects AI specifically across its client base to be a mid-single digit revenue generator. While AI-related activity is doing well, the demand for broader technology hardware remains weak, particularly in PCs and handsets (the latest forecast from Lenovo is for industry handset demand to fall 6% over 2023). For memory chip producers, despite the opportunity from advanced offerings, the current backdrop is extremely difficult with broader DRAM prices (including DDR4) expected to continue falling while consensus expects SK Hynix to report a loss for full year 2023.

However, on the flip side inventory across most of the chain has normalised back to reasonable levels with a much better H2 expected. For TSMC despite downgrades in forecasts for this year including after the recent Q2 results, analysts are forecasting Q4 results to show 20% quarter-on-quarter growth and a better 2024 as advanced nodes (on N3) are ramped up and an expectation of better performance from legacy nodes on seasonal peak and inventory rebuild. The commentary from Micron has a similar tone, stating the trough in revenue (for memory) is behind us and also pointing to a better upcoming cyclical environment (helped by significant industry capacity cuts) in addition to the positive secular drivers discussed above.

We have AI related exposure across foundry, memory, sensors, equipment, package and testing, motherboard and power related.

In any weakness and/or as we progress through the year, we are likely to be increasing our exposure across the spectrum in a sector with an improving cyclical outlook in conjunction with an ongoing powerful (positive) secular theme and which could prove supportive to the broader EM index for some time.