Erudite Asia

Hon Hai joins hands with Nvidia to bet on humanoid robot Liu Yangwei: focusing on Kaohsiung service applications Hon Hai Chairman Liu Yangwei announced a revenue target exceeding 7 trillion yuan for the year during the Kaohsiung Smart City Promotion Committee Annual Meeting. The company plans to collaborate with NVIDIA to invest in humanoid robot service applications in Kaohsiung, with plans to expand into Taipei and Keelung. The Kaohsiung City Government is focused on integrating AI technology with digital transformation to enhance smart city development and establish a competitive local AI industry chain. The meeting addressed key themes such as AI education, industrial innovation, and international market trends, featuring officials and industry leaders, including Mayor Chen Qimai and Liu Jingqing from the National Development Council. Yangwei indicated that the humanoid robot applications would use advanced hardware and Huida software technology, with intentions to partner with the city government to develop smart city software. Following sufficient public engagement with the app, Hon Hai aims to create a software development platform to produce localized mobile applications, ultimately striving for a self-sufficient AI ecosystem by merging software and hardware capabilities. #AI #Taiwan Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/gP6eBeAx

[AI Now] Will the AGI Era Begin in the New Year? OpenAI 'O3' Commercialization at a Crossroads Amid Expectations and Doubts OpenAI's new AI model, 'O3', has achieved record-breaking performance, becoming a leading contender in challenging AI tests and sparking discussions about its potential as a precursor to artificial general intelligence (AGI). What sets 'O3' apart from the 'GPT-4' series is its use of the 'Chain of Thought (CoT)' technique, which enhances problem-solving capabilities. CEO Sam Altman views 'O3' as the beginning of a next-generation AI that can tackle complex reasoning tasks. At the 'Shipmas' event, 'O3' excelled in the 'Arc AGI' benchmark, scoring 75.7% in low-performance mode and 87.5% in high-performance mode. It outperformed previous models, such as 'O1' and 'Claude 3.5', which scored only 32% and 53%, respectively. Francois Chollet, founder of Arc AGI, acknowledged this leap in task adaptability as significant. However, despite its remarkable results, skepticism exists among some experts regarding 'O3's implications for AGI. Critics, including Gary Marcus, argue that the 'Arc AGI' benchmark may not reflect real-world capabilities accurately and caution that 'O3's achievements could be inflated by heavy investment in benchmarking rather than practical utility. While 'O3' excels in programming tasks, it still struggles with the diverse challenges faced by human programmers. Furthermore, 'O3' faces the challenge of high operational costs, with optimal performance expenses reaching $6,000. Estimates for achieving its record scores could run into hundreds of thousands of dollars, raising concerns about large-scale commercialization feasibility. Although Altman remains hopeful about the timeline for AGI, he warns that its practical significance may not meet current expectations. #AI #SouthKorea Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/gQK3qSa3

Google and Nvidia are here! Big orders surge in foreign investment: MediaTek’s revenue will explode in the next two years Morgan Stanley's research highlights MediaTek’s successful entry into AI, driven by partnerships with tech giants like Google and Nvidia. These collaborations focus on AI PCs and application-specific integrated circuits (ASICs), positioning MediaTek to seize emerging AI opportunities. With increasing orders from Google, the company forecasts earnings per share to reach 123 yuan next year. Major mobile manufacturers, including Chinese brands, Samsung, and Apple, recognize MediaTek's technological expertise. The report points out MediaTek's competitive edge in cost control and execution, enabling it to rapidly gain market share in the ASIC sector, previously dominated by Broadcom and Marvell. The launch of a new AI PC chip with Nvidia is expected to boost sales, with projections of three million units annually. MediaTek’s collaborations with U.S. companies Apple, Google, and Nvidia aim to diversify its product lines and lessen reliance on the Chinese market. Revenue is anticipated to exceed 600 billion yuan this year and reach 798.2 billion yuan by 2027, highlighting substantial growth potential. Amid ongoing U.S.-China geopolitical tensions, MediaTek is well-positioned to benefit from both markets, leveraging its technological strengths to attract American clients while maintaining solid relations with Chinese customers. #Semiconductors #Taiwan Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/gJ7NvUtt

Great Southern New Silicon Valley launches investment promotion President: Semiconductors + AI build Taiwan’s resilience President Lai Ching-te has launched the "Greater Southern New Silicon Valley" project, focusing on the Tainan Sharon Smart Green Energy Science City to position Taiwan as a hub of artificial intelligence. This initiative aims to facilitate digital transformation across industries, enhance global competitiveness, and promote a balanced and healthy society. The National Science Council proposed the project, which aims to create a networked AI industrial ecosystem in Tainan, Chiayi, Kaohsiung, and Pingtung. Key developments include the "AI Innovation Application Building," set for completion in three months, alongside ongoing investment initiatives. Lai emphasized that success hinges on favorable timing, location, and local talent, particularly in Taiwan's robust semiconductor and AI sectors. Lai outlined plans for six major regions, with over one trillion yuan allocated for transportation, environmental improvements, and infrastructure. He highlighted the importance of integrated systems in advanced manufacturing, which dominates over 90% of the global market. The establishment of a national chip team and collaborations with firms like AMD and Huida further underscore initiatives to strengthen the AI and semiconductor industries, while local governments will optimize transportation, education, and environmental conditions to support AI industry growth. #Semiconductors #Taiwan Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/gr-rTYRH

The Taiwanese government is launching the Southern Taiwan Silicon Valley Program to establish the region as a hub for artificial intelligence (AI) development. President William Lai stressed the importance of leveraging Taiwan's semiconductor and machinery manufacturing strengths to provide comprehensive solutions across sectors like healthcare, infrastructure, and academia. This initiative aims to expand AI applications and bolster Taiwan's economic resilience amid global geopolitical shifts. National Science and Technology Council Minister Wu Cheng-wen noted that the program will cover science and industrial parks in southern cities, including Tainan, Kaohsiung, Chiayi, and Pingtung. Tainan is emerging as a semiconductor center, boosted by Advanced Micro Devices’ plan for a research-and-development center there. The program will promote collaboration between research institutes and high-tech industries, encouraging AI adoption among small and medium-sized enterprises. A key feature will be the Shalun Smart Green Energy Science City in Tainan, designed to integrate R&D facilities with commercial and residential spaces. Additionally, the initiative will enhance regional infrastructure and improve cultural, medical, and educational facilities to attract high-tech talent. #AI #Taiwan Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/gwFUY-X8

SK hynix will showcase its high-bandwidth memory (HBM) chips at CES 2025 in Las Vegas from January 7 to 10. The company will present its 16-layer HBM3e product, boasting a capacity of 48 gigabytes and the highest layered configuration. Key executives, including CEO Kwak Noh-jung and Chief Marketing Officer Kim Ju-seon, will attend to promote SK hynix as a comprehensive AI memory provider. The 16-layer HBM3e chip utilizes an advanced mass reflow-molded underfill process to enhance thermal performance and manage chip warpage. SK hynix expects the yield of this model to be on par with its 12-layer HBM3e chips in mass production. Additionally, the next-generation HBM4 chip is slated for mass production in the second half of this year, offering double the data transfer channels of HBM3e. Beyond HBM products, SK hynix will also showcase high-capacity enterprise SSDs (eSSDs), including the 122 terabyte D5-P5336 from Solidigm, which optimizes power and space efficiency. Other highlights include LPCAMM2 and ZUFS 4.0 memory solutions designed for on-device AI applications. The company aims to emphasize its advances in AI memory technology with a diverse range of tailored solutions at CES 2025. #Semiconductors #SouthKorea Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/gjXe6iNX

CES 2025 in Las Vegas introduces a wealth of innovations themed "Dive In." The event features advancements in AI, mobility, digital health, and home entertainment, with a new focus on quantum technology in collaboration with the World Quantum Congress, highlighting its ability to enhance AI deployment. Approximately 4,800 companies from 160 countries are participating, marking a record turnout from Korea. AI remains a central focus, showcasing practical applications in productivity, customer experience, and healthcare. Samsung Electronics CEO Han Jong-hees discusses AI-driven living, while LG presents AI-enhanced appliances, including the Smart InstaView Refrigerator, which checks food freshness. Samsung’s booth will also display smart home devices powered by the Bixby AI assistant. SK Telecom and its affiliates highlight infrastructure for AI data processing, showcasing products like advanced AI memory chips from SK hynix. Discussions between SK Group Chairman Chey Tae-won and Nvidia CEO Jensen Huang are expected. The competition in television technology is evident, with both Samsung and LG introducing 100-inch Mini LED TVs amid pressure from Chinese competitors. In mobility, tech companies and traditional automakers reveal new innovations. LG's Mobility eXperience platform aims to personalize vehicle interiors with AI, while Hyundai Mobis showcases a holographic windshield display. Toyota provides updates on its Woven City prototype. Additionally, Samsung’s C-Lab features fifteen AI-focused startups, complemented by LG NOVA and Naver’s venture capital initiatives promoting health and AI advancements. #AI #SouthKorea Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/gFfskzmd

The Japanese semiconductor industry is poised for revival, driven by soaring chip demand from generative artificial intelligence and rising geopolitical tensions. Key player Rapidus Corp. is advancing with plans to launch a pilot line and new production facilities this year. In December, Rapidus installed extreme ultraviolet (EUV) lithography equipment at its Chitose, Hokkaido plant, a pivotal technology for advanced semiconductor production. CEO Atsuyoshi Koike highlighted this milestone as crucial for positioning Japan as a global supplier of cutting-edge semiconductors. The company aims to initiate mass production of 2-nanometer chips by 2027, with the pilot line expected to start operations in April. #Semiconductors #Japan Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/g373dMXt

Popular Concept Stocks》Special Application IC (ASIC) Demand for application-specific integrated circuits (ASICs) is surging, driven by the booming artificial intelligence (AI) sector. Companies are increasingly developing AI ASICs to offer alternatives to traditional general-purpose chips. Unlike CPUs and GPUs, ASICs are tailored for specific tasks, making them ideal for training and inference in AI. The market is estimated to grow from $12 billion in 2024 to $30 billion by 2027. Taiwanese manufacturers stand to gain from this trend. Shixin-KY has reported significant revenue growth, with 75% of its income stemming from ASIC design. Other firms like Creative and Faraday are also adapting; Creative focuses on System on Chip (SoC) design, while Faraday is advancing its process technology to the 2nm node. Despite some revenue dips, these companies continue to innovate. MediaTek is investing heavily in ASIC development, targeting a market potential of $45 billion by 2028. Its strategy includes strengthening partnerships with cloud service providers, positioning the company for substantial growth in the ASIC sector. The shift toward AI reflects a broader trend among manufacturers adjusting operations to meet the increasing demand for customized semiconductor solutions. #Semiconductors #Taiwan Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/g5z4KJsi

Is the Blackwell Effect coming? Morgan Stanley: CES has two highlights, RTX 4090 is in short supply Morgan Stanley Securities' latest report on the "AI Supply Chain" highlights key developments ahead of the U.S. Consumer Electronics Show (CES), focusing on game graphics processors and AI PCs. The NVIDIA RTX 4090 is expected to remain in short supply as production ramps up for the RTX 5090, which utilizes the new Blackwell architecture. Taiwanese supply chains, particularly TSMC, KYEC, and ASE, are poised to benefit from these trends. The ongoing demand for graphics processing is sustaining the RTX 4090 shortage. With the lifecycle of mainland China's H20 product nearing its end, Blackwell GPU shipments are projected to exceed initial forecasts this quarter. Taiwanese packaging and testing facilities are optimizing operations, supported by KYEC's significant capital investments slated for 2025. Analyst Zhan Jiahong predicts a dramatic rise in AI GPU shipments, potentially exceeding 900,000 units by Q1 2025. The CES event may showcase Huida's WoA AI PC chip, built on advanced TSMC 3nm technology, though prior software compatibility issues may delay its launch until Computex. Meanwhile, MediaTek's Arm SoC is forecasted to generate around US$2 billion in revenue by 2025. Morgan Stanley maintains an optimistic outlook for Taiwanese graphics card and AI PC supply chains, assigning "outperform" ratings to several companies. Target prices are set for key players including TSMC, MediaTek, KYEC, ASE Investment Holdings, Asustek, and Gigabyte, indicating a positive market outlook as CES approaches. #Semiconductors #Taiwan Want more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/g-bEck7a