In this podcast, the Ninety One Quality team analyses the semiconductor industry through a quality investment lens and discusses which companies may be the long term winners in the space.
It is no exaggeration to say that modern life wouldn’t really be possible without them. Also known as semis, or chips, semiconductors can be found in thousands of products such as computers, smartphones, appliances, gaming hardware, and medical equipment. These chips – which can act as an electrical conductor or insulator – really are the brains of all modern electronics, with those at the cutting edge having a diameter on par with a DNA molecule. For context, the global semiconductor market generated around US$600 billion in revenue in 2021, with some predicting the industry to top US$1 trillion by 20301.
We break up this complex space into three: the end markets, the types of chips, the value chain approach. The end market is where the semiconductors end up, with smartphones accounting for about one-quarter of the US$600 billion. PC’s and servers are sizeable markets too (~US$100 billion), with cars another important – and accelerating – market as the world transitions towards electric vehicles, which need about double the semiconductors that traditional cars do.
There are three main types of chips: logic, memory and analog. Within logic, you have a central processing unit (CPU) – which handles the core processing tasks – and a graphic processing unit (GPU) – which handles complex, data-intensive tasks. Memory chips – defined as either DRAM or NAND – store data, which the logic chips take and use accordingly. Analog chips are the bridge between the real world and the digital world, taking a real-world sense, such as sound or light, and turning it into a digital signal.
The value chain involves five types of companies. The designers – such as Nvidia or Qualcomm – spend their entire time designing chips. Manufacturers – like TSMC – make the chips. Integrated device manufacturers (IDMs) – such as Samsung or Intel – do both. Electronic design automation (EDA) developers – including Cadence and Synopsys – create software for the designers, while capital equipment providers produce the machines used to make the chips, with ASML and Lam Research two examples.
Semiconductors do have very attractive structural growth trends – evidenced by the 2030 estimate – but there is inherent cyclicality and capital intensity embedded in the space. Therefore, we believe it is important to tread carefully in the sector and focus on companies that can be more insulated from that inherent cyclical risk.
In Asia, for instance, there is Samsung, which is the market leader in the memory industry followed by Hynix and Micron, and these three dominate an industry was very fragmented just 10-15 years ago. The memory industry remains the most cyclical element of the semiconductor space, with supply imbalances relatively common following spikes in demand and pricing – as we are seeing now. Yet Hynix and Micron have announced capex reductions of 50%, whereas Samsung is maintaining its capex, meaning they should continue to gain market share. The result of this is their margins have remained relatively stable.
Over in Europe we have ASML, a Dutch semiconductor capital equipment provider that produces the machines that the likes of Samsung uses for memory chips or TSMC for logic chips. They focus on the lithography step of the chip-making process, which is almost like a high-tech photocopier. Generating and harnessing extreme ultraviolet light (EUV) is an almost unfathomably difficult engineering challenge – one that proved too demanding for competitors Canon and Nikon to even attempt. ASML’s EUV machines – about the size of a single-decker bus, at a cost of US$150 million – have become precisely what chip makers need to print smaller circuits while increasing capacity and speed. Therefore, in the EUV era, ASML is no longer just a dominant supplier, but rather a monopoly, created through technological brilliance and innovation, as opposed to the more traditional M&A route. ASML’s importance in the supply chain has therefore become entrenched, evidenced by its record order book stretching out to 2028.
The US is the world leader in terms of design, but the manufacturing of semiconductors has largely been outsourced to Asia. Because of its strategic importance to everyday life and recent supply chain issues – especially in the autos space – politicians are getting involved. China has made chip independence a national priority in its latest five-year plan, unveiled in March 2021, while the Biden administration is effectively continuing his predecessor’s policies, which included blacklisting more than 60 companies deemed a threat to national security, restricting their access to US components. More recently, the CHIPS Act of 2022 unveiled a US$280 billion package to boost US semiconductor capacity, innovation and national security. There are geographical considerations too; tensions between the US and China over Taiwan – which is home to TSMC, the world’s largest chipmaker, - remain fraught.
Not necessarily. For context, the CHIPS Act includes about US$50 billion in semiconductor subsidies. Yet if we take the five biggest companies in terms of capex spend (Intel, Samsung, TSMC, SK Hynix and Micron), they have committed to spend US$600 billion over the next five years. While subsidies generate noise and headlines, and short-term disruption, we believe business fundamentals remain the key driver for the sector; this should be the focus of long-term investors.
Companies are also able to mitigate the headwinds, too. For instance, Nvidia is still allowed to sell slightly less powerful chips to the Chinese market, and ASML – with the Dutch government under pressure to follow US rules – is almost customer ambivalent; they do not mind where chips are made given their status as an infrastructure provider to the industry. If there is a reallocation of capacity, ASML machines will still be used. Samsung is seeking to diversify its production base, opening a factory in Arizona in 2024.
There is always going to be some level of disruption coming through from politics or new disruptors coming through, but the many decades of capital requirements and knowledge needed to shift the dial in this space means that we believe the economic moat – or competitive advantages – around these companies is secure for the long-term.
1Source: McKinsey, April 2022.
No representation is being made that any investment will or is likely to achieve profits or losses similar to those achieved in the past, or that significant losses will be avoided.
This is not a buy, sell or hold recommendation for any particular security.
To listen to this and other podcasts, subscribe on Spotify, Apple Podcast, Google Podcasts or simply search for ‘Ninety One Big Picture’ on your podcast app of choice.