How AI Is Pushing the Semiconductor Supply Chain to the Limit | Bloomberg Primer

The semiconductor industry, the engine of the global economy, is experiencing unprecedented growth, driven largely by the artificial intelligence boom. This industry, which produces trillions of semiconductor devices annually, is projected to reach $1 trillion in revenue by 2026. The smallest components on modern chips are incredibly minute, often smaller than viruses and tens of thousands of times narrower than a human hair. These sophisticated chips can command prices of up to $30,000 and are crucial for everything from smartphones and AI servers to defense systems. The manufacturing process for these chips is considered the most complex on Earth, pushing the boundaries of software, hardware, chemistry, and physics. A critical bottleneck in this process is lithography, the technique used to transfer intricate designs onto silicon wafers. ASML, a European tech firm, holds a dominant position in this specialized field, producing the highly complex lithography machines that are indispensable for manufacturing advanced chips. All leading chip manufacturers rely on ASML's technology. To create progressively smaller and more intricate chip features, ASML developed Extreme Ultraviolet (EUV) lithography, a technology that generates a type of light not naturally found on Earth. These EUV machines are enormous, weighing as much as a blue whale and costing $400 million each. They enable the printing of smaller resolutions by using larger lenses and achieve high speeds with rapidly moving stages capable of extreme acceleration. ASML's machines are the sole source for the world's most advanced lithography tools, making them a hyper-specialized and crucial link in the concentrated semiconductor supply chain. Companies like AMD are at the forefront of designing the semiconductor products that drive leading-edge computing. AMD produces server chips for supercomputers, networking chips, and chips for gaming consoles, all of which require significant compute capabilities and energy efficiency. The fundamental building blocks of these chips are transistors, acting as nanoscopic switches that control electrical flow and generate the binary code powering digital information. Chips are ubiquitous, found in virtually every electronic device, from phones to coffee makers. The stock market has reacted enthusiastically to the surge in chip demand, with AMD, in particular, leveraging the AI chip market into a multi-billion dollar business. A silicon wafer is the base material upon which hundreds of individual chips are printed. After processing, these wafers are cut into individual chips, a process that gives the term "chip" its origin. Chips range from basic analog chips with simple functions to advanced CPUs and GPUs, and highly specialized AI chips designed for rapid data processing. The silicon required for AI chips is significantly larger and more complex, often involving stacked chiplets and memory layers. AI chips are a rapidly growing segment of the market, accounting for over a quarter of all chips sold in 2025 and projected to exceed half by 2029. This demand has shifted the semiconductor industry from a cyclical model to a period of sustained growth, largely due to the build-out of data centers for AI applications. Spending on AI infrastructure, including data centers, is expected to surpass $1 trillion. However, the semiconductor supply chain faces significant concentration risks. Taiwan Semiconductor Manufacturing Company (TSMC), based in Taiwan, manufactures over 90% of the world's most advanced chips. This concentration in Taiwan, a region with geopolitical instability, presents a vulnerability. The global supply chain is intricate, with chip design primarily occurring in the US, raw wafers in Japan and Taiwan, machinery from the Netherlands, and components sourced globally. All these elements converge in fabrication facilities, or "fabs," heavily concentrated in Taiwan. TSMC's role as the world's largest contract chip manufacturer makes it a critical bottleneck. Building new fabs is a complex and experienced-driven process, leading to the clustering of facilities. This clustering, while efficient, also makes the supply chain precarious, susceptible to disruptions from natural disasters like earthquakes, pandemics, and geopolitical tensions. The standoff between China and Taiwan, with China claiming Taiwan as its territory, adds a layer of significant risk, with potential economic repercussions estimated in the trillions of dollars if conflict were to arise. In response to these vulnerabilities, global superpowers are investing heavily in domestic chip manufacturing. The US government, through the CHIPS Act, is incentivizing the reshoring of semiconductor manufacturing. While the US once manufactured nearly 40% of the world's chips, this figure has dropped to around 10%. Efforts are underway to increase this percentage, despite higher labor and regulatory costs, through subsidies and incentives. China, aiming for technological self-reliance and supremacy, is also aggressively developing its domestic chip industry. Despite facing US sanctions that previously hindered companies like Huawei, China has made significant strides, particularly in analog chip production. The country has invested billions in chip funds and private sector support to bolster its capabilities. Reports indicate secret manufacturing efforts and a surprising debut of a China-made advanced chip in 2023, demonstrating rapid progress. However, China's advanced chip-making has been hampered by fluctuating US regulations on access to Western technology, including ASML's lithography machines. The US is also seeing major expansion efforts, notably by Texas Instruments (TI), a company known for its foundational chips. TI, which produces chips for a vast array of devices, is investing $60 billion in advanced manufacturing. They are transitioning to larger 300-millimeter wafers, which can yield significantly more chips per wafer, offering a competitive advantage. TI's new mega-site in Texas, the size of two Empire State buildings, is designed to produce hundreds of millions of chips daily. While TI does not focus on AI chips, they anticipate that the growth in the AI sector will drive demand for their essential, lower-end chips, which are crucial for building AI infrastructure. The semiconductor industry is undergoing a profound transformation. Chips are no longer just behind-the-scenes components; they now define critical infrastructure, communications, mobility, defense, and AI. The global demand for semiconductors, fueled by AI and our increasingly technologically dependent world, is prompting a worldwide re-evaluation of manufacturing and supply chains. Companies are focusing on innovation, new product development, and cost savings to navigate this evolving landscape. The industry, once characterized by brutal economics, is now experiencing renewed excitement and a redefined perception, with chips themselves becoming recognized as "sexy."