AI will have different impacts on companies in the semiconductor sector
The rise of artificial intelligence was unexpected and has driven an explosion in interest across cloud service providers, enterprises and consumers. Investors are still working out the long-term implications; the MFS® global technology sector team are unpacking what AI means for different business models and how it will impact companies, whether, for example, it will be disruptive or positive for software business models.
In the semiconductor sector, there are many companies benefiting from the training of large language models, such as chip manufacturers, data centers and software designers. We are confident AI will be long-term positive for semiconductor businesses and are looking for companies underappreciated by the market. There is a wide range of outcomes from the perspective of units, prices, market share and gross margins as new competition will emerge to rival existing players. Being anchored in valuation helps us stay away from the AI hype cycle while understanding each of these businesses helps us uncover the opportunities and risks.
Geopolitical tensions affect companies, but supply chains are highly interconnected
Deglobalization risks have been one of the biggest focus areas of the global sector team for several years. Given the supply chain, semiconductors are one of the most globalized industries: Semiconductor design companies are centralized in the United States, capital equipment companies are in the US, Netherlands and Japan, manufacturers are largely out of Taiwan while the chemicals used during manufacturing come from all over the world. This is because the semiconductor sector lends itself to small niches, such as the design of manufacturing equipment or process technology.
The highly focused, niche-based model has led to a diverse but interconnected supply chain, making it challenging for any of these businesses to pull out of the system. As investors, we need to analyze their financial statements and understand the impact of geopolitical tensions on metrics like units sold and market share. We also need to know the extent to which companies manufacture exclusively in one location or region, how much it would cost them to diversify and whether there is enough capacity elsewhere to appropriately diversify. It is more difficult for companies using more advanced technology to diversify than it is for those using older chips. We are consistently talking with companies and assessing the potential impact on units, prices, margins and capex across each of the businesses, as well the exposure in our portfolios.
Energy usage is important from sustainability and business model perspectives
Energy is vital to semiconductor-related companies. For example, cloud service providers developing large language models are constrained by their energy usage. This makes energy consumption a sustainability focus as well as critical to their business models. A prime example of the impact is the increasing demand for Nvidia’s chips. Nvidia’s graphic processing units are more powerful and efficient than traditional central processing units as their GPU accelerates a single workload, making the process faster and using much less energy than a CPU.
We’re seeing semiconductor companies focusing on energy, trying to reduce their usage and managing energy flow efficiently through the computer system. From an investment perspective, we are assessing energy from both a sustainability and business model perspective.