Financial Services Association, September 25 — On Wednesday (September 25) local time, well-known consulting giant Bain released a report saying that the surge in market demand for artificial intelligence (AI) semiconductors and AI-enabled smartphones and laptops may lead to the next global chip shortage.

As we all know, the last global semiconductor shortage occurred during the COVID-19 pandemic, when supply chains were disrupted and people were forced to work from home, leading to a marked increase in consumer demand for electronic products.

Bain & Company wrote in the report that just like the demand caused by the pandemic, the upcoming wave of AI devices may also accelerate the purchase of smartphone and personal computer upgrades. In addition to the trend of upgrades, Bain also pointed out that AI-driven graphics processing units (GPUs) alone are sufficient to increase total demand for certain upstream components by 30% or more.

Source: Bain & Company website

Currently, tech giants are snapping up GPUs for data centers, most of which come from$NVIDIA (NVDA.US)$company. Earlier this month, Nvidia CEO Hwang In-hoon mentioned that demand for Nvidia products is very strong. “Everyone wants to be the first to receive products, and everyone wants to receive the most products.”

At the same time,$Qualcomm (QCOM.US)$Other companies are designing chips for smartphones and personal computers. The goal is to enable these devices to run AI applications locally without connecting to the cloud via the Internet, Samsung,$Microsoft (MSFT.US)$Other companies have already launched such products.

Bain said that the trend of GPU and AI consumer electronics products may be the main reason for future chip shortages. Coupled with ongoing geopolitical tension and other supply risks, it may eventually trigger the next round of semiconductor shortages.

Anne Hoecker, head of Bain's American technology and cloud services business, told the media, “The surge in demand for GPUs will cause shortages in some elements of the semiconductor value chain. If demand for CPUs is combined with the demand for AI-enabled devices, it may speed up the renewal cycle of PC products, which may lead to wider restrictions on semiconductor supply.”

Bain pointed out that the semiconductor supply chain “is extremely complex, and an increase in demand of about 20% or more is likely to break the balance, leading to a shortage of chips.” According to the report, “The explosive growth of AI at the intersection of large terminal markets may easily exceed this threshold, leading to weak bottlenecks in the supply chain.”

Take GPU as an example. The GPU designed by Nvidia requires$Taiwan Semiconductor (TSM.US)$For OEM production, TSMC also relies on manufacturing tools and materials from many countries around the world during production, such as the Netherlands$ASML Holding (ASML.US)$The lithography machine. Huang Renxun also mentioned that Nvidia is very dependent on TSMC, and transferring orders to other suppliers may cause chip quality to deteriorate.

Bain wrote at the end of the report that leading companies will learn from the recent chip shortage and maintain a safe balance between inventory shortages and surpluses. They will sign long-term purchase agreements to secure chips and manufacturing capacity for some time to come.

The report also analyzed that more companies will design and use industry-standard semiconductors as much as possible rather than application-specific chips, and they will continue to invest in supply chain resilience against geopolitical uncertainty.

edit/emily