According to a new report from Taiwan Economic Daily, TSMC has increased the orders of machines required for CoWoS (Chip on Wafer on Substrate) packaging by 30%. The company has reportedly taken this step as the demand for semiconductors to be manufactured using this technology has increased drastically from its clients Amazon, AMD, Broadcom, and Nvidia.
The usage of Generative AI models, such as Bard, ChatGPT, and Dall-E, is skyrocketing. These technologies require AI chips to work, such as AMD Instinct and Nvidia H100, and these semiconductors are manufactured using the CoWoS packaging technology offered by TSMC. Since the usage of GenAI is increasing drastically, the demand for chips required to run this technology is also growing rapidly.
As a result, Amazon, AMD, Broadcom, and Nvidia, who get their AI chips manufactured by TSMC, are placing rush orders to get as many AI chips as they can as fast as possible to meet the increasing demand for GenAI. However, TSMC doesn’t have as many CoWoS packaging machines as it requires to meet the current demand, putting its facilities under strain and forcing it to order more machines.
TSMC to manufacture 30,000 wafers every month based on CoWoS packaging technology from Q2 2024
With the current number of CoWoS packaging machines, TSMC reportedly manufactures around 12,000 wafers monthly. The company ordered more CoWoS machines in May this year to increase the production capacity to about 15,000 to 20,000 wafers per month. However, looking at an even higher demand, TSMC has now increased the orders for machines by 30%, and with that, the company is expected to manufacture around 30,000 wafers per month.
TSMC has reportedly contacted multiple local companies, including AllRing-Tech, E&R Engineering Corporation, Grand Process Technology, Group Up Industrial, and Scientech Corporation, to procure additional CoWoS packaging machines. The machines TSMC ordered in May are expected to be installed by Q1 2024, and the ones it has ordered now are expected to be installed by Q2 2024.
