The demand for TSMC’s CoWoS advanced packaging is skyrocketing. Following NVIDIA’s expansion confirmation in October, there are reports in the industry that major clients like Apple, AMD, Broadcom, Marvell, and others are also placing additional orders with TSMC.

To meet the demands of these five major clients, TSMC is fast-tracking the expansion of CoWoS advanced packaging capacity. Next year, the monthly capacity will increase by about 20% more than the original doubling target, reaching 35,000 wafers, reported by UDN News.

TSMC has not commented on the capacity deployment for CoWoS advanced packaging. Industry sources believe that the substantial orders from TSMC’s major clients indicate a widespread growth in AI applications, driving the demand for chips such as GPU and AI accelerators.

In response to the continuous increase in AI demand, TSMC had previously announced the doubling of CoWoS advanced packaging capacity expansion for next year but did not disclose the monthly production capacity. Industry reports suggest that TSMC’s CoWoS advanced packaging capacity next year will not only double but will also increase by an additional 20% from the original target, resulting in a total monthly capacity of 35,000 wafers.

NVIDIA currently stands as the main large customer for TSMC’s CoWoS advanced packaging, securing almost 60% of TSMC’s related capacity, which is used in its AI chips such as H100 and A100. Additionally, AMD’s latest AI chip products are in the mass production stage, and the upcoming MI300 chip, expected to launch next year, will adopt both SoIC and CoWoS advanced packaging.

At the same time, Xilinx, a subsidiary of AMD, has been a significant customer for TSMC’s CoWoS advanced packaging. With the continuous growth in AI demand, not only Xilinx but also Broadcom has started increasing orders for TSMC’s CoWoS advanced packaging capacity.

(Image: TSMC)

Explore more

• [News] TSMC Deploys Manpower to Support Longtan and Tainan Facilities Amid CoWoS and 3nm Demand
• Exploring the Significance of 3D-SOC and 3D-IC in Cutting-Edge 3D Advanced Packaging

NVIDIA and Intel are adapting to the latest U.S. chip export restrictions by introducing downgraded AI chips specifically tailored for the Chinese market, UDN News said.

According to insider from the China Star Market, a Chinese media, NVIDIA has developed three downgraded AI chip models for the Chinese market. Intel also plans to release downgraded Gaudi 2 chip with an aim to US restriction.

NVIDIA’s latest downgraded AI chips, including HGX H20, L20 PCle, and L2 PCle, are anticipated to be unveiled after November 16th. Chinese companies are likely to receive samples in the coming days. These three chips, derived from the modification of NVIDIA H100, will align their performance with parameters below the new U.S. regulations. Ongoing communication with NVIDIA suggests mass production is slated for the year-end, said by industry sources.

Besides, Yicai also confirms from multiple NVIDIA supply chain sources. The three AI chip products are designed for cloud training, cloud inference, and edge inference, with specific launch times pending confirmation. Sampling is projected between November and December this year, followed by mass production from December this year to January next year.

On the Intel front, there are rumors of a response plan. As reported by The Paper, Intel is planning to release an improved version of its Gaudi 2 chip. Although the rumor exists, specific details are yet to be disclosed.

Since the U.S. government introduced chip export control to China last year, NVIDIA initially designed downgraded AI chips A800 and H800 for Chinese companies. However, new regulations in October this year by the U.S. Department of Commerce brought A800, H800, L40S, and other chips under control. Failure to secure export permission may necessitate order cancellations for NVIDIA.

(Image: Nvidia)

On October 17, 2023, the U.S. government once again expanded its restrictions on the export of semiconductor devices and products to China. The newly added control conditions now encompass NVIDIA’s L40S, A100, H100H800, as well as general-purpose AI server GPUs tailored for the Chinese market, such as A800 and H800. Additionally, AMD’s MI200 series, MI300 series GPUs, and Intel’s Habana Labs’ Gaudi 2, Gaudi 3 GPUs fall under the regulatory framework.

Recalling the U.S. government’s export restrictions on AI chips issued to IC design firms in September 2022, at that time, only A100, H100, and MI200 series were subjected to control, and the U.S. Department of Commerce granted NVIDIA and AMD a one-year buffer period.

In contrast, the recent regulations not only cover all mainstream AI server GPUs but also eliminate the buffer period for these chip companies. In essence, companies or institutions in countries not permitted for export can only acquire AI server chips with performance potentially inferior to NVIDIA L40S or AMD MI200 series for the next few years.

Furthermore, stricter control thresholds for lithography equipment have led to the inclusion of ASML’s DUV, the 1980Di, in the control list. This equipment is primarily used in the 28 ~ 7nm process. Previously controlled products were focused on the EUV 3000 series for 7nm and below processes and the DUV 2000 series for 16/14 ~ 5nm processes.

This move indicates that the U.S. government’s desire to control semiconductor process technology has officially extended to mature processes of 28nm.

The expanded U.S. controls on AI chips and semiconductor manufacturing devices not only target China but also countries that might collaborate with Chinese institutions and businesses in AI development.

In this scenario, China is left with only two viable options to establish efficient AI computing resources: (1) designing and mass-producing AI server chips itself or (2) utilizing the computing resources of cloud service providers.

As the U.S. is also discussing the potential inclusion of cloud service providers in semiconductor control policies and currently formulating relevant countermeasures, this path remains unreliable for China. Therefore, the only dependable option is to independently design and manufacture AI server chips.

Read more

• [News] Nvidia says US speeded up new export restrictions on AI chips
• [News] Chinese Chip Equipment Rises Amid U.S. Restrictions

According to CTEE, NVIDIA’s forthcoming AI server, the GB200 (B100), slated for a 2024 release, has entered the certification phase in the supply chain. Recent market rumors suggest that Foxconn, originally intended to secure orders for the B100 board, faced certification challenges. As a result, Wistron has maintained its initial order share.

Additionally, it is worth noting that Ingrasys, a subsidiary of Foxconn, is actively manufacturing the H100 product and is a strong contender to secure orders.

Unofficial sources indicate that NVIDIA initially considered making Foxconn the second supplier for AI-GPU server baseboard in the upcoming B100 series. However, due to yield concerns and other factors, Wistron is still expected to receive 100% of the orders. Wistron has also capitalized on the opportunity to secure orders for the front-end AI-GPU module, which appears to be a successful move.

The rapid evolution of AI has intensified competition among assembly plants. Wistron and Foxconn play crucial roles as suppliers for NVIDIA’s current mainstream H100 series GPU modules and baseboards.

Wistron, as the exclusive supplier for H100 baseboards in the NVIDIA DGX and HGX architectures, also holds the exclusive role of providing mainboards and assembling AI servers for DGX. As shipments of the H100 series AI servers, built on the NVIDIA DGX and HGX frameworks, steadily increase in the latter half of the year, Wistron’s AI server-related product business shows consistent growth.

It’s worth noting that Ingrasys is responsible for manufacturing the H100. NVIDIA’s founder, Jensen Huang, and Foxconn’s Chairman, Young Liu, jointly attended a technology event, highlighted the close collaboration in between, underscoring Foxconn’s determination to secure B100 orders.

(Image: Wistron)

At this year’s Qualcomm Snapdragon Summit, the company announced its latest PC processor, the Snapdragon X Elite. With impressive performance metrics, this development is poised to shake up the PC processor market as Arm architecture gains ground, posing a substantial challenge to the established x86 architecture.

At this year’s Qualcomm Snapdragon Summit, the company announced its latest PC processor, the Snapdragon X Elite. The launch of laptops featuring the Qualcomm Snapdragon X Elite is expected in mid-2024, marking an opportune moment for a “counteroffensive.”

TrendForce indicates that Arm architecture PC processors have secured around an 11% market share this year, primarily propped up by Apple’s laptop processors. Industry insiders reveal that, in light of the growth potential in the PC processor market, semiconductor giants are increasingly adopting ARM architecture to venture into the market.

2024 Sees Laptop Upgrade Surge, Desktop Market Shrinks                  

Statistics reveal that the surge in remote work during 2020 prompted a shift in consumer preferences from desktop computers to laptops. Moreover, the ongoing establishment of cloud platforms by businesses in 2021 and 2022 has generated positive momentum, signaling a shrinking desktop market and an expanding PC market.

AI-powered PCs and Windows 12 next year are expected to ride a fresh wave of upgrades in 2024. Therefore, when PCs featuring ARM architecture become widespread, Intel and AMD may not be predominantly affected in the laptop processor business based on the x86 architecture. Instead, the desktop processor segment could face the most significant impact.

Kedar Kondap, Qualcomm’s Senior Vice President and General Manager of the Compute and Games Division, foresees an upgrade wave fueled by AI PCs next year, with further growth anticipated in 2025. It is expected that consumers will lean towards AI PCs for their next computer purchases.

The initial wave of products equipped with Qualcomm’s AI PC processors has been unveiled, aligning with the upcoming wave of device upgrades in next year. While Intel is set to launch its first AI acceleration engine, the Intel Core Ultra, featuring integrated NPU in December, its Microsoft Windows 12 certification remains a point of observation.

In a broader perspective, Intel and AMD are positioned to follow up with the AI PC trend by 2025. This coincides with the ending service of Windows 10 and the gradual implementation of Wifi 7 and 6G technologies. By 2028, they are expected to play a pivotal role in driving AI PC growth.

On another note, a South Korean analyst anticipates that the growth momentum in AI PCs hinges on when Apple incorporates AI features into Mac computers.

ARM vs. x86, Microsoft’s Crucial Role                                    

This is because Microsoft is set to launch Windows 12 next year, featuring the built-in Copilot AI assistant. It will collaborate with operating systems and software such as Windows, Edge, Microsoft 365, Outlook, and the Bing search engine, ushering in an entirely new AI-driven user experience.

Several tech giants are fiercely competing in the AI PC market, with NVIDIA and AMD investing in the development of Arm architecture processors. It’s worth mentioning that in 2016, Microsoft agreed to let Qualcomm exclusively develop Windows-compatible chips, and this agreement is set to expire in 2024. Consequently, Qualcomm may gain a strategic advantage. In contrast, the collaboration between NVIDIA and MediaTek on Arm processors might only begin to bear fruit in 2025.

As for AMD’s foray into Arm architecture research and development, whether this indicates a less optimistic outlook for the x86 market is a matter for ongoing observation. Intel CEO Pat Gelsinger expressed that he isn’t concerned about Arm architecture processors vying in the PC market. From a different perspective, Intel may even consider assisting with manufacturing.

(Image: Qualcomm)