In the sudden wake of DeepSeek, many of us seem to have already forgotten the importance of AI infrastructure. Admitting DeepSeek has brought a whole new concept of AI model as well as compute and energy, which inevitably raised concerns about whether investors have overvalued the whole AI industry.  At the same time, however, with the vast amount of AI usage, one question remains, can the new AI model architecture support the overwhelming volume of data usage?  

What we could see is the new AI model indeed brought us into a new era of AI, but that is not to say that what we have invested previously was wasted. We believe that while the volatility of the moment might overshadow the future, the building blocks of AI remain strong, and its foundation is only to grow stronger. The companies that power the AI ecosystem, starting with raw materials and manufacturing equipment, are critical to this advancement.

The AI ecosystem is far from just a collection of smart algorithms and machine learning models. Beneath the surface, there exists a robust infrastructure built on semiconductors, memory chips, and the machinery that makes them all possible.  

This weekly piece, marking the first write-up of a sequel, will explore the pivotal role played by companies like Lam Research (LRCX US), Applied Materials (AMAT US), and Taiwan Semiconductor Manufacturing Company (TSM US; 2330 TW) in shaping the AI sector’s growth. Despite recent market volatility, these companies’ essential contributions to the production of semiconductors and advanced equipment continue to drive the AI industry forward.

The Foundation of AI: Raw Materials and Equipment

At the heart of the AI revolution lies one critical component: semiconductors. They are the brainpower that drives every aspect of AI, from deep learning to natural language processing. But to understand how AI is formulated and operates, to fully understand AI and managing volatility, we must first look at the foundational stages of its development, which rely heavily on semiconductor manufacturing, particularly wafer fabrication.

LRCX and AMAT, two of the most prominent players in this field, design and manufacture equipment essential for semiconductor production, particularly focusing on processes like etching, deposition, and cleaning. These stages are critical in shaping the raw materials into the sophisticated chips that power AI systems. In short, we are well aware of what TSMC does; LRCX and AMAT, on the other hand, provide the equipment and materials that companies like TSMC use for semiconductor manufacturing (foundry).

Etching and Deposition: Transforming Raw Materials

LRCX and AMAT are leaders in wafer fabrication. Their machines enable the deposition and etching processes that are fundamental to semiconductor production. Etching involves removing material from the wafer surface to create intricate patterns, while deposition involves layering material onto the wafer to build complex circuit structures.

These technical processes are central to producing semiconductors that enable the most advanced AI systems and are major process to gather a complete semiconductor. While companies like ASML play a pivotal role in the patterning stage using light, LRCX and AMAT handle the groundwork, making their contributions just as critical. They work in tandem with companies like ASML, whose technology enables the critical patterning of wafers but does not perform the foundational work of transforming raw materials into semiconductor-ready chips.

The Growing Demand for Memory Chips

As AI continues to scale, one component that plays a vital role is high-bandwidth memory (HBM), which is essential for AI model training and large-scale data processing. The HBM market is projected to grow exponentially, by over 10x in the next decade driven by the demand for more powerful AI systems. Both LRCX and AMAT are fully integrated into the supply chain, as their wafer fabrication equipment is essential in the production of the memory chips that will power AI innovation.

Client Base and Market Impact

The major players in the HBM market are key customers for LRCX and AMAT. For instance:

• LRCX serves top clients like Samsung, Micron Technology (MU US), Intel (INTC US), TSM, and SK Hynix.
• AMAT similarly counts Samsung, TSM, INTC, and MU among its largest clients.

These companies’ financials remain strong, with both LRCX and AMAT posting consistent margins and earnings. In 2024, both companies have maintained net margins no less than 26%, with an average EBITDA margin of around 30%. This demonstrates their stable growth in a sector with high demand for AI-related components.  

Valuation  

As mentioned, LRCX and AMAT are cornerstone companies in the AI industry. However, because they provide services at the first stage of AI development, their stock prices are primarily driven by the major players in the industry, such as NVDA, TSMC, AVGO, QCOM, and others.

Nonetheless, the correlation with these giant players has not pushed up their valuations in the same way it has for other companies in the sector. LRCX P/E at 24.77x, AMAT at 21.08x, while NVDA at 50.65x, AVGO at 87.75x, and AMD at 94.99x. Furthermore, LRCX and AMAT are supported by strong financial fundamentals, with LRCX's EV/Sales at 6.52x and AMAT's at 5.36x, compared to NVDA at 22.57x and AVGO at 22.18x. These figures suggest that LRCX and AMAT are undervalued in relative terms, while NVDA may be overvalued, as the stock market is driven by expectations. That said, in the era of AI, it is important to pay attention to the fundamentals of the sector as well.

‍

TSM: The Backbone of AI

While LRCX and AMAT provide the equipment for semiconductor production, TSM represents the critical manufacturing arm that transforms the raw materials into finished products. TSM is the world’s leading semiconductor foundry, responsible for manufacturing chips designed by other companies, including many of the biggest players in AI.

As the world’s largest semiconductor manufacturer, TSM holds a dominant position in the market, producing chips for global giants like Apple, NVIDIA, and Qualcomm. In the context of AI, TSM’s role is indispensable. Without TSM’s advanced manufacturing capabilities, many of the AI advancements we see today, from powerful GPUs (NVDA A200) to memory chips, would not be possible.

‍

Why They Matter in the AI Revolution ‍

In short, companies like LRCX, AMAT, and TSM are not just supporting the semiconductor industry, they are laying the groundwork for AI’s future. Without the innovative equipment produced by LRCX and AMAT, and the advanced chip manufacturing provided by TSM, AI would not have the scale or capability it does today.

These companies are at the forefront of building the AI ecosystem. From raw materials to finished semiconductors, their technologies are enabling the development of smarter, more efficient AI models that continue to evolve.

As the AI industry grows, so does the demand for the equipment and materials these companies provide.

Despite the challenges brought on by DeepSeek and other market disruptions, the AI sector's potential remains vast, and the fundamental which drives this growth are well-positioned to capitalize on the expanding market.