Smart AI Investing Isn’t About Chasing the Big One

The future impact of artificial intelligence (AI) remains uncertain, with ongoing discussions about the potential for achieving artificial general intelligence (AGI) and the capabilities of current neural-network systems, like those powering chatbots such as ChatGPT. While some investors may be tempted to speculate on future breakthroughs, a more prudent approach involves focusing on established investment strategies. A key area to consider is the intricate supply chain that supports AI technology, which is often riddled with bottlenecks. At these critical junctures, certain companies possess significant economic advantages, enabling them to generate substantial profits more quickly than the ambitious developers of large language models. A pivotal example is found in the chip design sector, where demand for products from Nvidia Corp. is so intense that OpenAI has recently sought alternative solutions, securing a major agreement with competitor Advanced Micro Devices Inc. for its inference functions. The landscape of AI investment is also witnessing a data center expansion, particularly in the United States. Recent data shows that construction spending in this sector surged to an annualized rate of $41 billion, accounting for nearly half of all office-related expenditures. However, developers face challenges related to sourcing chips and securing reliable power generation. For instance, acquiring gas turbines, essential for energy production, has become increasingly difficult, with lead times extending from two years to five or more. This delay is largely due to a market dominated by a few key players, including GE Vernova Inc., Siemens Energy AG, and Mitsubishi Heavy Industries Ltd., who are cautious about expanding capacity in light of fluctuating demand. Moreover, the high barrier to entry in turbine manufacturing deters many startups from entering the market, given that refining turbine designs can take decades. Another noteworthy segment is the market for high-bandwidth memory (HBM), crucial for supporting AI systems. South Korea’s SK Hynix Inc. leads this sector, followed closely by Samsung Electronics Co. and Micron Technology Inc. The technical complexities have even led Huawei Technologies Co. to rely on Korean products for some of its AI processors, despite a preference for domestic sourcing. OpenAI has also forged strategic partnerships with Samsung and Hynix to secure its supply chain ahead of competitors, resulting in strong stock performance for these companies this year. However, pinpointing the leading companies in the AI space is not as straightforward as analyzing market dynamics and entry barriers. For instance, ASML Holding NV, which has a near-monopoly on lithographic equipment for high-performance chips, has not seen its stock price rise in line with the AI boom, primarily due to its concentrated customer base and export restrictions affecting sales to China. Investors without a technical background may find it easier to identify supply chain challenges rather than speculate on the potential for AI to achieve human-like intelligence. Historical trends from the smartphone era can provide valuable insights into supplier dynamics. Additionally, the AI-driven stock market surge is not confined to the US or China, as the complexity of the supply chain means that no single country can dominate all aspects of AI development. Concerns about an AI bubble are valid, especially as large tech firms experience significant daily fluctuations in stock prices. However, the varied performance of companies like gas turbine manufacturers and ASML suggests that there are still rational investment opportunities amidst the current rally. Ultimately, the majority of profits in the AI sector are likely to be realized upstream, where the most significant supply chain bottlenecks exist. By 2029, as OpenAI approaches profitability, traditional engineering firms in the turbine sector may have already reaped considerable financial rewards.