Nvidia Secures Over 70% of TSMC's Advanced Packaging Capacity This Year

Nvidia, the leading AI hardware and software company, has secured over 70% of TSMC's advanced packaging capacity for this year, according to sources cited by the Economic Daily News. This strategic move underscores Nvidia's dominance in the AI chip market and its commitment to driving innovation in high-performance computing (HPC) and AI applications.



TSMC, the world's largest dedicated independent (pure-play) semiconductor foundry, has been fully committed to placing orders for advanced processes and advanced packaging to meet the substantial order demands from major cloud service companies. The high demand for AI chips from global cloud service providers, such as Amazon AWS, Microsoft, Google, and Meta, has led to a fierce competition among these companies to bolster their AI server arsenal. This, in turn, has resulted in a supply shortage for AI chips from major manufacturers like Nvidia and AMD.

TSMC anticipates robust growth in AI applications, with revenue from server AI processors expected to more than double this year. These processors are projected to contribute a low-teens percentage of the company's total revenue in 2024, with a 50% compound annual growth rate (CAGR) over the next five years. By 2028, server AI processors are expected to contribute over 20% to TSMC's revenue.

Nvidia's next-generation Blackwell architecture AI chips, including the B100, B200, and the GB200 with Grace CPU, are expected to significantly enhance computational capabilities compared to the H100 series. These chips utilize TSMC's 4-nanometer process with an enhanced version known as N4P, along with higher-capacity and updated specifications of HBM3e high-bandwidth memory.

On the other hand, AMD's MI300 series AI accelerators are manufactured using TSMC's 5-nanometer and 6-nanometer processes. AMD integrates CPU and GPU dies vertically using TSMC's SoIC advanced packaging before employing CoWoS advanced packaging with HBM. This method introduces an additional step of advanced packaging complexity with the SoIC process.

TSMC is actively expanding its advanced packaging capacity to meet the growing demand from customers like Nvidia. By the end of this year, TSMC's CoWoS monthly capacity is expected to reach between 45,000 to 50,000 wafers, and SoIC monthly capacity is forecast to reach 5,000 to 6,000 wafers. This expansion will enable TSMC to support Nvidia's increased capacity and maintain its market share in the AI chip sector.

In conclusion, Nvidia's dominance in TSMC's advanced packaging capacity, particularly for CoWoS and SoIC technologies, has significant implications for the competitive landscape for AI chip manufacturers like AMD. The supply constraints, increased competition for resources, potential supply shortages, TSMC's focus on Nvidia's needs, and the potential impact on AMD's product roadmap all contribute to a challenging environment for AMD to compete effectively in the AI chip market. However, TSMC's expansion of advanced packaging capacity and Nvidia's increased capacity for AI chips have significant implications for TSMC's revenue growth and market share in the AI chip sector. TSMC's ability to support Nvidia's expanded capacity, while also maintaining its market leadership and potentially capturing additional market share from other AI chip manufacturers, will be crucial in driving its revenue growth and market share in the AI chip sector.