The supply chain for chips is fragile and may interrupt AI/ML projects. The situation will not change without immediate domestic chip production funding.
Key findings from a Department of Commerce report commissioned last year indicated that the global semiconductor chip shortage would persist without action on the part of the government. Many of the findings from the report proved true, and the situation has gotten worse due to continued supply chain issues and more recent disruptions due to the fighting in Ukraine.
The report, Risks in the Semiconductor Supply Chain request for information (RFI), noted first that the median inventory held is less than five days, way down from 40 days in 2019. With razor-thin margins, any disruption could shut down factories and cause major economic disruption. As the Biden administration considers funding domestic chip production to the tune of $52 billion, experts believe that making the move now is critical.
- A 17% higher demand for chips in 2021 than in 2019, with no increase in supply during the intervening years.
- Confirmation that most manufacturing facilities are operating at full or near full capacity, correcting the idea that the shortage lies with manufacturing.
- An understanding of where bottlenecks happen—legacy logic chips and analog chips are on this list. It will continue to cause challenges in auto and medical device manufacturing, radio, and power management.
- Discovering that fab capacity is the main identified bottleneck, followed by a lack of raw material inputs for conductors and related components.
The Department of Commerce will continue to work with the private sector to find ways to ease the chip shortage as new disruptions emerge. The proposed plan would allow economic sectors to avoid shutting down and ease supply chain constraints. The report includes responses from all parts of the supply chain and contains responses from more than 150 sources.
A look to the future and the impact of a chip shortage on AI
The chip shortage is expected to continue through 2022 and into 2023. One aspect to consider is that gaining consistent, reliable, and compressed time to insights and business outcomes using AI/ML requires investments in purpose-built and right-sized infrastructure. If the shortage persists, AI project may find they do not have the processing power to achieve success.