Navigating supply chain dynamics for sustained AI growth

The rapid rise of artificial intelligence (AI) has transformed industries, with supply chain management emerging as a critical application area. However, the infrastructure supporting AI growth, particularly the supply chain for AI algorithm training and inference, faces significant challenges. This Chapter examines the complexities of the AI supply chain, focusing on hardware and energy requirements.
The AI supply chain involves hyperscale data centers equipped with advanced GPUs and CPUs, along with lithography and advanced packaging processes to produce these components. The demand for AI-centric semiconductors is projected to grow substantially, with estimates indicating a need for an additional 130-240 GW of data center capacity over the next seven years. This growth is driven by the increasing deployment of AI inference, which requires robust and efficient hardware.
However, the supply chain faces bottlenecks, including long lead times for chip production and significant energy demands. Grid access and energy production are critical constraints that could slow data center growth and keep AI inference costs high. The risk of bullwhip and pork cycle effects further complicates the supply chain, likely leading to periods of overcapacity and unstable pricing for multiple years.