Jefferies' global head of equity strategy, Chris Wood, continues to advocate for a "picks and shovels" investment approach in the artificial intelligence (AI) sector. This strategy centers on identifying companies that provide the foundational infrastructure and raw materials necessary for the AI industry's growth, rather than investing directly in the developers of AI models. This mirrors the historical gold rush, where providers of tools like Levi Strauss and companies selling mining equipment often profited more consistently than the gold miners themselves.
Key areas of investment under this strategy include semiconductors, memory, networking, power, and cooling solutions. For instance, semiconductor manufacturers like Taiwan Semiconductor, Samsung, and Intel, along with equipment suppliers such as ASML, Lam Research, Tokyo Electron, and Applied Materials, are considered beneficiaries. Advanced Micro Devices (AMD) and Nvidia develop Graphics Processing Units (GPUs) essential for parallel processing in generative AI, while companies like Broadcom and Marvell produce Application-Specific Integrated Circuits (ASICs) crucial for AI inference work. Intel and AMD also produce CPUs that manage AI agents and provide storage.
Memory chips, supplied by Micron, SK Hynix, and Samsung, are also a critical bottleneck, facing shortages due to limited production capacity. Networking solutions, including copper and optical interconnects from companies like Amphenol, Lumentum, and Coherent, are vital for fast communication within AI data centers. Furthermore, the massive power demands of data centers create opportunities for manufacturers of power generation equipment, including those specializing in renewable energy, gas turbines, and fuel cells. The significant heat generated by these facilities drives demand for liquid cooling systems from companies like Vertiv and Schneider.
Beyond direct technology, the physical construction and maintenance of data centers present further "picks and shovels" opportunities. This includes suppliers of steel, concrete, and raw materials such as copper, with companies like Anglo American and Freeport McMoRan benefiting. Copper prices, for example, reached a new intraday record above $14,600 per metric ton in early September, partly driven by AI data center buildouts. Jefferies has also initiated coverage on critical minerals stocks, highlighting demand driven by AI, with Element Solutions, Almonty Industries, IperionX, and Materion Corporation receiving "Buy" ratings. Materion, for instance, saw its stock deliver a 113% return over the past year and reported second-quarter adjusted earnings of $1.90 per share, exceeding forecasts.
Wood's perspective is that these established businesses often have diverse revenue streams and may offer more attractive valuations compared to some AI model developers. While acknowledging risks, such as new technologies displacing older solutions, Jefferies believes the unprecedented capital investment boom in AI outweighs these risks, leading to rapidly growing revenues for hyperscalers and providing significant "picks and shovels" opportunities.
