Artificial Intelligence has become the hot topic of the year, having expanded rapidly in many fields and captured global attention. The language around it swings from the hyperbole of true believers to the nay saying of doomsday prophets. This excitement is evident in the meteoric rise of AI-related stock prices. Beyond the excitement, however, lies a looming challenge: explosive demand for computing power is likely to hit supply shortages in the next two years.
The “ex” driven by AI can be seen in many fields. In sectors like healthcare, AI tools have enhanced diagnostics and personalized treatments, allowing for faster and more accurate decision-making. Manufacturing automation is driving efficiencies, reducing production time and cutting costs, while technologies such as generative AI and machine learning allow companies to optimize supply chains, predict maintenance needs and implement flexible production methods. Furthermore, AI has transformed customer service by streamlining operations and significantly enhancing the customer experience. AI-driven chatbots and virtual assistants, such as those used by Amazon, Bank of America and Microsoft, can handle vast volumes of inquiries, providing customers with instant responses to common questions, order tracking, or technical support without human intervention.
On a broader economic scale, AI’s impact is poised to be massive. McKinsey estimates that by 2030, AI could contribute an additional $13 trillion to the global economy, equivalent to a 16% boost to global GDP. This growth is being driven by increased productivity through labor automation, innovations, and new competitive dynamics across multiple industries.
Investors have viewed this expansion favorably. American Century Investors reported that strong demand for AI-related technology stocks fueled the surge in U.S. stocks during the first half of 2023. Four companies—Microsoft, Nvidia, Amazon, and Meta—accounted for two-thirds of the S&P 500 Index’s return during this period.
AI applications require advanced semiconductors, such as graphics processing units (GPUs), to process large datasets and power sophisticated models like generative AI. To understand the future opportunities in the AI industry, it’s essential to understand the basics of semiconductors—what they are, how they’re made, and who makes them.
GPUs require a wide variety of materials such as copper, cobalt, and tungsten, as well as multi-level operations.
Nvidia controls 80% of the global GPU market, sourcing nearly all its chips from Taiwan. These GPUs, critical for training AI models (like OpenAI’s ChatGPT), and major technology companies continue to acquire GPUs from Nvidia. These GPUs are typically located in data centers—a physical location with servers and computing machines. As Nvidia notes in a blog post on their website, GPUs are exceptionally well-suited for AI tasks due to their ability to handle parallel processing. Unlike traditional CPUs, which are designed for sequential operations, GPUs can perform thousands of tasks simultaneously, making them ideal for training AI models that require immense computational power.
The AI chip market also encompasses chips for smartphones and computers that allow these devices to run AI applications locally—as in the case of Apple’s new AI-equipped smartphones. Companies like Qualcomm are leading the endeavor to develop this application. Notably, Qualcomm also sources nearly all its chips from Taiwan. Other big tech players, however, have started to design their own custom AI Chip in the U.S—attempting to reduce cost and reliance on Nvidia and Qualcomm. For example, Microsoft has developed its own custom AI chip ,Azure Maia, which is specifically designed to train large language models, while Google has developed “Tensor Processing Units” (TPUs) to handle advanced AI calculations.
According to Bain & Company, the increased demand for both GPUs and AI-powered consumer electronics is likely to contribute to a growing chip shortage, putting pressure on the supply chain and potentially limiting the availability of these crucial components. Bain and Company estimates that “the AI-driven surge in demand for graphics processing units alone could increase total demand for certain upstream components by 30% or more by 2026.”
This supply-demand imbalance is contributing to price increases across several industries that rely on semiconductors, including automotive, consumer electronics, and manufacturing. The automotive industry, for example, has already experienced significant production delays due to the lack of critical chips. As Bain & Company’s report notes, as the shortage intensifies, industries that depend on AI applications—such as e-commerce and healthcare—may face reduced operational efficiency and higher costs. Consumer products like smartphones and laptops are also expected to see price hikes as chip manufacturers struggle to meet rising demand for AI-enabled devices.
To address this challenge, companies are pursuing strategies like signing long-term contracts and diversifying their supply chains to include suppliers in regions beyond Taiwan. Governments are also stepping up. For example, the United States passed the CHIPS Act of 2022, which invests almost 53 billion dollars in funding to increase semiconductor production in the United States—creating jobs and supporting innovation.
As Fieldston’s computer science teacher, Mr. Kurt Vega explains, “The big problem is because chips are offshore, mainly in Taiwan…The CHIPS Act I think is a very good initiative, but it takes years and billions of dollars to get these things up and running. So there could absolutely be a chip crunch.” His insight highlights the precarious dependence on international supply chains for semiconductors and the significant time and investment required to bolster domestic production. Mr. Vega highlights the urgency of looming the supply chain challenge in the swiftly growing A.I. sector.
