It is understandable that large U.S. technology companies like Nvidia and Microsoft have become favorites in artificial intelligence (AI), but investors looking to take advantage of advances in machine learning must pay close attention to emerging markets.
Since the launch of ChatGPT in late 2022, investors have flocked to the stocks of AI-related companies trading in the U.S. These tech companies, the main beneficiaries, have the technical expertise, experience, and financial resources to turn the new technology into a huge commercial success. In fact, Nvidia’s stocks, which manufacture the chips that train computers for AI, have skyrocketed by 250% and giants like Microsoft and Meta have experienced dizzying increases. But investors cannot be complacent with valuations. After all, these stock prices incorporate a lot of future growth.
Those looking to profit from the development of the AI revolution need to look beyond the U.S. technology sector, considering that alternatives abound. Among the overlooked are a group of crucial Asian companies in the AI supply chain. Specifically, the most promising source of profitability may come from around 40 listed companies that are involved in activities in Taiwan, South Korea, and China. They manufacture nearly all the chips for AI in the world and for many essential products assisted by AI.
First, there are companies that are part of the supply chain for graphic processing units (GPUs). All of Nvidia’s AI processors are manufactured, packaged, and integrated by companies based in Taiwan. In fact, as Nvidia increases the processing power and memory of AI chips, it increasingly depends on a sophisticated processor packaging technology known as Chip on Wafer on Substrate (CoWoS), perfected by TSMC, Taiwan’s largest semiconductor group and global chip manufacturer – the U.S. government will provide up to $11.6 billion in aid to manufacture in the country-.
Another source of opportunities lies in the high prices of Nvidia processors, with opportunities for cheaper AI chip manufacturers aimed at supplying Nvidia’s U.S. competitors, such as Microsoft or Amazon’s chip unit. Among the most promising options are Application-Specific Integrated Circuits (ASICs). They can be used in the development of AI models trained to infer results from real-time data analysis with costs 10% to 20% lower than Nvidia’s AI servers and with shorter delivery times. This is the case for specialized companies like the Taiwanese Alchip and Wiwynn.
In addition, Asian hardware manufacturers assisted by AI. Indeed, the new generation of smartphones and computers enhanced with integrated AI imply a new cycle of technology personal substitution, which can greatly benefit these companies. According to some analysts, the compound annual growth rate of sales of phones units with integrated AI can be approximately 80% up to 2027.
So it is expected that this year alone TSMC’s AI-related revenue will double and exceed 10% of its total sales, and that AI integrated application integrated circuits will represent 30% of the sales of Wiwynn, its competitor also based in Taiwan. Even SK Hynix, the second largest company by market capitalization in South Korea, leveraging the demand for high-bandwidth memory chips, key in AI infrastructure, has surpassed the milestone of $100 billion in market capitalization. It has started production of a new version used by Nvidia. Goldman Sachs estimates that the high-bandwidth memory chip segment will reach $23 billion by 2026.
The expected twelve-month forward price-to-earnings ratio of AI supply chain companies in emerging markets is around 19 times, a considerable discount compared to U.S. semiconductor companies, which have traded at 27 times earnings, considering that the annual growth forecast of these emerging market companies is 14% and 26% in terms of revenue and profits.
