SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : ASML Holding NV
ASML 1,067+0.7%Nov 3 9:30 AM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
From: BeenRetired5/8/2024 10:00:24 AM
   of 42273
 
TSMC "packaging capacity is booked out for two years"

TSMC packaging booked out
by NICK FARRELL
on07 MAY 2024



We can’t fit you in for a couple of years

TSMC has declared that its sophisticated packaging capacity is booked out for two years after Nvidia, AMD, and Guanghuida* have secured its state-of-the-art packaging technologies for their high-performance computing projects.

The focus on high-performance computing, which plays a crucial role in driving artificial intelligence tasks, is a key factor in TSMC's growth strategy. The company anticipates a significant boost in revenue from AI processors, with projections indicating a doubling this year alone. Over the next five years, the annual growth rate for AI chips is expected to hit fifty per cent, with AI processors projected to contribute over twenty per cent of TSMC's revenue by 2028.

Both Nvidia and AMD have used TSMC's advanced packaging capacities for their products. Nvidia's H100 chip, developed using TSMC's 4nm process, uses Chip-on-Wafer-on-Substrate (CoWoS) packaging. In contrast, AMD's MI300 series, manufactured with TSMC's 5nm and 6nm processes, employs System-on-Integrated-Chip (SoIC) for CPU and GPU integration before applying CoWoS with High Bandwidth Memory.

Guanghuida, an emerging name in the AI chip market, has reserved TSMC's packaging capacity. Their H100 chips, powered by TSMC's 4nm process and CoWoS packaging, feature SK Hynix's High Bandwidth Memory for improved performance. Guanghuida's latest Blackwell architecture AI chip, based on TSMC's advanced 4nm process, includes upgraded HBM3e memory, which doubles the computing power compared to earlier models.

The increasing demand for AI chips is driven by major global cloud service providers such as Amazon, AWS, Microsoft, Google, and Meta, which are competing for dominance in the AI server market. With supply shortages from key manufacturers like Nvidia, AMD, and Guanghuida, these cloud giants are turning to TSMC to meet their needs, contributing to the chipmaker's positive revenue forecasts.

To address this growing demand, TSMC is expanding its production capacity for advanced packaging. By the end of this year, the monthly production of CoWoS is expected to triple, reaching between 45,000 to 50,000 wafers, while the capacity for SoIC is set to double, achieving 5,000 to 6,000 wafers. By 2025, the monthly production of SoIC is projected to double once more, reaching 10,000 wafers.

The full booking of TSMC's advanced packaging capacity highlights the rapid innovation in AI-driven computing, with major players strategically preparing to take advantage of this growing market.

*Copilot:
Response stopped

Guanghuida is a rising player in the AI chip market 1 2. They have booked TSMC’s packaging capacity for their H100 chips, which are powered by TSMC’s 4nm process and CoWoS packaging 1 2. These chips feature SK Hynix’s High Bandwidth Memory (HBM) for improved performance 1 2.

Moreover, Guanghuida’s latest Blackwell architecture AI chip, based on TSMC’s advanced 4nm process, includes upgraded HBM3e memory, which doubles the computing power compared to earlier models 1 2. This indicates that Guanghuida is heavily invested in the development and production of high-performance computing chips, particularly for AI applications 1 2.

Learn more

1gizmochina.com 2fudzilla.com 3fudz.fudzilla.com
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext