Original source (on modern site) | Article images: [1]
Taiwan's artificial intelligence (AI) supply chain is evolving from an original equipment manufacturing role into technology integration, driven by global technology giants stationing top-tier executives in Taiwan to accelerate technology development. As AI technology advances, collaboration across diverse ecosystems is crucial because the complexity and societal impact of modern AI exceed what any individual enterprise, academic institution or single discipline can achieve. "Collaboration" dominated the discourse at technology forums during the Semicon Taiwan trade show in Taipei last week. The consensus among many keynote speakers and senior technology executives is clear: The future of AI hinges not on isolated breakthroughs, but on radical collaboration. While Silicon Valley tech giants such as Nvidia Corp craft the digital blueprints of AI, Taiwan turns these visions into reality through its world-class hardware manufacturing. With the acceleration of AI, cross-industry collaboration has become essential to mitigate risks early in the development process, where a single mistake can be financially devastating. Such localization could help shrink time to market. To that end, Google plans to expand its Taiwan office space by 60 percent as local partners take on a growing role in developing the company's AI infrastructure, Google senior vice president and chief technologist for AI and infrastructure Amin Vahdat, said last week. Taiwanese chip designer MediaTek Inc helps design Google's tensor processing unit, a custom application-specific integrated circuit to accelerate AI machine learning workloads, as the US technology giant seeks to reduce its dependence on Nvidia for AI chip supply and to cut the cost of ownership. Carl Zeiss AG last week said it is expanding its local engineering team and assigning more technology and application engineers to Taiwan to promote "knowledge exchange." The German company is ASML's sole supplier of its extreme ultraviolet lithography optical systems, a critical component in chipmaking. It is a key technology used to make the world's most advanced semiconductor chips, such as 2-nanometer chips. Zeiss is developing an advanced inspection tool for next-generation copackaged optics technology, which is crucial to overcoming bandwidth and thermal issues in AI data centers. Zeiss is considering bringing this new optical inspection tool to Taiwan for on-site testing. The company is also planning to build a semiconductor innovation center in Taiwan, which would minimize the company's research-and-development team's international travel, it said. Merck Group, one of the world's largest semiconductor material suppliers, is replicating this strategy to improve customer proximity and minimize service latency. Bing Han (韓冰), executive vice president of its global thin films business, is moving from California to Taiwan next month, highlighting the company's decision to move senior technical and decisionmaking talent forward to Asia. The arrangement is designed to shorten the distance between technology development, customer validation and decisionmaking, Merck said, adding that Taiwan's innovation and research investment would only grow. Han said he has visited Taiwan more than 50 times in the past few years. As the number of companies globally capable of manufacturing the most advanced semiconductors is limited, and materials suppliers seeking to participate in cutting-edge process development cannot simply deliver products after receiving specifications — they must engage with customers earlier to jointly solve process challenges, Merck said, adding that speed, customer-centricity and enhanced internal collaboration are the key elements in the company's latest restructuring drives. These technology giants' latest movies indicate that Taiwan is increasingly shifting toward a "heterogeneous" integration of AI technology, rather than simply being a manufacturing hub of AI hardware. As US AI chip supplier Nvidia and memory chipmaker Micron Technology Inc have built their presence in Taiwan and deepened on-site collaboration to integrate wafers, memory chips and cooling systems directly within local facilities for assembly and testing, it triggers a new shift of "anchoring localization" in Taiwan. Such on-site collaboration could be the optimal solution for the AI ecosystems to catch up with rapid AI chip migration and explosive growth of new AI models.