According to UDN’s report, the Taiwanese government has introduced the “Taiwan Chip-Driven Industrial Innovation Program,” an investment of NT$300 billion with an annual average of NT$30 billion.  On October 2, Executive Yuan Commissioner Tsung Tsong Wu and Chairperson of the National Science Council revealed that this program welcomes foreign participation but will consider prerequisites such as competition with local talent.

When asked about what attracts international startups to Taiwan, Tsung Tsong Wu emphasized that while the $30 billion investment over a decade is significant, Taiwan’s strength lies in its complete industry ecosystem, covering everything from IC design and manufacturing to packaging and testing.

The government and private sector collaborate to achieve the best results, making international startups and chip designers eager to come to Taiwan. The aim is to turn Taiwan into the global destination for related industries’ dreams.

The National Science Council presented the “Taiwan Chip-Driven Industrial Innovation Program” to the Executive Yuan on October 2. Tsung Tsong Wu explained that out of next year’s $12 billion technology budget, approximately $4 billion will be allocated to generating AI-driven innovations across all industries, attracting international investments, and supporting startups. Additionally, around $8 billion will be dedicated to strengthening talent development and advancing towards cutting-edge processes.

While the initial investment for the first year of the “Taiwan Chip-Driven Industrial Innovation Program” is $1.2 billion, critics have raised concerns about its insufficiency. The National Science Council clarified that although the initial investment is relatively low, the budget allocated will increase from 2025 onwards, adapting to the evolving economic landscape.

Tsung Tsong Wu emphasized that the program encompasses both advanced and mature processes. While Taiwan is perceived to have an advantage in advanced processes, many mature processes, such as 3DIC, are vital for future industry innovations.

(Photo Credit: TSMC)

The US Department of Commerce issued new restrictions on AI chips on October 17, 2023, with a focus on controlling the export of chips to China, including NIVIDA’s A800, H800, L40S, and RTX4090, among others. Taiwanese manufacturers primarily serve cloud service providers and brand owners in North America, with relatively fewer shipments to Chinese servers. However, Chinese manufacturers, having already faced two chip restrictions imposed by the US, recognize the significance of AI chips in server applications and are expected to accelerate their in-house chip development processes.

TrendForce’s Insights:

1. Limited Impact on Taiwanese Manufacturers in Shipping AI Servers with H100 GPUs

Major Taiwanese server manufacturering companies, including Foxconn, Quanta, Inventec, GIGABYTE, and Wiwynn, provide AI servers equipped with H100 GPUs to cloud data centers and brand owners in Europe and the United States. These Taiwanese companies have established some AI server factories outside China, in countries such as the US, the Czech Republic, Mexico, Malaysia, and Thailand, focusing on producing L10 server units and L11 cabinets in proximity to end-users. This strategy aligns with the strategic needs of US cloud providers and brand owners for global server product deployment.

On the other hand, including MiTAC, Wistron, and Inventec, also provide server assembly services for Chinese brands such as Inspur and Lenovo. Although MiTAC has a significant share in assembling Inspur’s servers, it acquired Intel DSG (Data Center Solutions Group) business in July 2023. Therefore, the focus of AI servers remains on brand manufacturers using H100 GPUs, including Twitter, Dell, AWS, and European cloud service provider OVH. It is speculated that the production ratio of brand servers will be adjusted before the new restrictions are enforced.

Wistron is a major supplier for NVIDIA’s AI server modules, DGX A100, and HGX H100. Its primary shipments are to end-users in Europe and the United States. It is expected that there will be adjustments in the proportion of shipments to Chinese servers following the implementation of the restrictions.

Compal has fewer AI server orders compared to other Taiwanese manufacturers. It has not yet manifested any noticeable changes in Lenovo server assembly proportions. The full extent of the impact will only become more apparent after the enforcement of the ban.

During the transitional period before the implementation of the chip ban in the United States, the server supply chain can still adapt shipments based on local chip demand in China to address market impacts resulting from subsequent chip controls.

2. Chinese Manufacturers Focusing on Accelerating In-House Chip Development

Chinese cloud companies had already started developing their AI chips before the first U.S. chip restrictions in 2022. This included self-developed AI chips like Alibaba Cloud’s T-HEAD, a data center AI chip, and they expanded investments in areas such as DRAM, AI chips, and semiconductors with the aim of establishing a comprehensive IoT system from chips to the cloud.

Baidu Cloud, on the other hand, accelerated the development of its third-generation self-developed Kunlun chip, designed for cloud and edge computing, with plans for an early 2024 release.

Tencent introduced three self-developed chips in 2021, including an AI inference chip called Zixiao, used for Tencent’s meeting business; a video transcoding chip called Canghai, used in cloud gaming and live streaming applications; and a smart network card chip named Xuanling, applied in network storage and computing.

ByteDance made investments in cloud AI chips through its MooreThread initiative in 2022 for applications in AI servers. Huawei released the Ascend 900 chip in 2019 and is expected to introduce the Ascend 930B AI chip in the latter half of 2024. While this chip has the same computational power as the NVIDIA A100 chip, its performance still requires product validation, and it is speculated that it may not replace the current use of NVIDIA GPUs in Chinese AI servers.

Despite the acceleration of self-developed chip development among Chinese cloud server manufacturers, the high technological threshold, lengthy development cycles, and high costs associated with GPU development often delay the introduction of new server products. Therefore, Chinese cloud companies and brand manufacturers continue to purchase NVIDIA GPUs for the production of mid to high-end servers to align with their economic scale and production efficiency.

In response to the new U.S. restrictions, Chinese cloud companies have adopted short-term measures such as increasing imports of existing NVIDIA chips and building up stockpiles before the enforcement of the new restrictions. They are also focusing on medium to long-term strategies, including accelerating resource integration and shortening development timelines to expedite GPU chip manufacturing processes, thus reducing dependency on U.S. restrictions.

In the wake of a semiconductor shortage, Chinese automakers have veered onto the path of self-developed chips over the past two years. Recently, Changan Automobile, in collaboration with the Chongqing High-Tech Industrial Development Zone and the Intelligent Manufacturing Industrial Research Institute, established Chongqing Xinlian Integrated Circuit Co., Ltd. This venture, boasting a considerable registered capital of 8.7 billion yuan, signifies a substantial investment from Chongqing’s state-owned entities and major automobile manufacturers. It is dedicated to advanced production of 12-inch large-scale integrated circuits.

Changan is not alone in this endeavor; companies like Geely, GAC, BAIC, BYD, and others have embarked on self-development plans or have chosen to enter the chip manufacturing domain through partnerships. Emerging forces in the automotive industry like XPeng, NIO, and Li Auto are also opting for self-developed chips.

The Rise of Self-Developed Chips

Tesla stands as the pioneering automaker in developing its self-driving chips. Industry insiders suggest that their decision was fueled by the inadequacy of chip suppliers like NVIDIA and the ample funds generated from Tesla’s surging sales. Their approach has been widely recognized by the market, prompting others to explore this direction.

In the realm of self-developed chips, different car manufacturers adopt diverse strategies. Companies like Tesla, XPeng, and NIO, renowned for their self-developed algorithms, focus on high-performance chips.

An industry source emphasized that car manufacturers prefer to stress full-stack self-development, but off-the-shelf chips cannot fully leverage the advantages of self-developed algorithms. Thus, powerful companies opt for customized chips to align with their proprietary algorithms. This underscores the need for automakers to possess robust capabilities in autonomous driving software and algorithms.

Notably, NIO has assembled a 300-member chip team, focusing on self-driving and LiDAR chips. XPeng’s chip team is developing high-computing power self-driving chips similar to Tesla’s FSD chip. Furthermore, Li Auto expanded its chip team and collaborated with Sanan Optoelectronics to establish a power semiconductor production line in Suzhou.

In contrast, traditional domestic auto manufacturers often commence their self-developed chip ventures with power semiconductors due to their higher onboard usage and relatively lower development complexity. Several carmakers have partnered with chip companies for mass production collaborations. Horizon Robotics, for instance, has signed mass production agreements with mainstream auto manufacturers like BYD, Great Wall, Li Auto, and Changan.

(Photo credit: Changan Automobile)

A small subtropical island off the coast of southeast China, Taiwan is subject to certain cyclical weather changes throughout the year, the most notorious of which is its yearly typhoons that at different times benefit its agricultural industry and cause various natural disasters.

Much like everything else that happened in 2020, last year marked a stark exception for the island’s climate, which saw no typhoons, resulting in a relatively dry year. Compounding the issue is the current season of low precipitation. Taken together, these factors have since resulted in a significant drought that required an all-hands-on-deck approach from the government, such as rationing water on specific weekdays and advising industries to cut down on water consumption.

Meanwhile, a similar drought has been taking place in the global semiconductor world. As the arrival of the COVID-19 pandemic last year brought fundamental changes to the way we work, study, and live, so too has the general public’s consumption of electronic devices – and, in turn, the worldwide demand for chips used in these devices – risen.

If there are bumps in the road to Taiwanese foundries’ continued dominance, lack of rain certainly isn’t one.

Seeing as how Taiwan is the central hub of the world’s advanced semiconductor technologies, acts as home to industry leader TSMC (which is the exclusive supplier of Apple’s M1 processors), and accounts for more than half of the world’s chip manufacturing capacity, industries and media alike are fearing that the domestic drought will exacerbate the current global chip shortage, since chip fabrication processes require enormous amounts of clean water.

However, true to its market leadership, the Taiwanese semiconductor industry has so far remained unaffected, at least on the supply side, by the water shortage. This is in part due to the fact that domestic foundries (i.e., chip manufacturers) have previously completed numerous drills related to worst-case scenarios of long droughts and are accordingly well prepared in these extenuating circumstances. Furthermore, the foundries also signed contracts with utility companies to ensure an ample supply of water to keep fabs (semiconductor fabrication plants) running via water tank trucks.

TrendForce therefore expects domestic foundry operations to continue unabated for the time being. Case in point, on April 15, TSMC announced an increased capital expenditure of US$30 billion for 2021. The foundry is also actively expanding its production capacity of mature technology processes in response to the growing demand from clients worldwide.

In the world of semiconductors, advancements in process technologies occupy merely one part of the equation when it comes to long-term success. Other requirements pertain to governmental, infrastructural, climate, procedural, and talent-related dimensions, just to name a few.

While Taiwanese foundries look for a way out of the ongoing drought, they are not only acing these requirements in spades, but also staying in the spotlight of the electronics supply chain in light of geopolitical tensions, oligopolistic market trends, and the persistent global health crisis. If there are bumps in the road to Taiwanese foundries’ continued dominance, lack of rain certainly isn’t one.

（Cover image source：TrendForce）