Single Incident Will Not Affect AI Demand
NVIDIA CEO Jensen Huang recently admitted to an extramarital affair with businesswoman Huang Le, who is suspected by U.S. intelligence agencies of having ties to China, raising U.S. national security concerns. NVIDIA's board has launched an independent investigation, and the U.S. Congress has also asked Jensen Huang to attend a closed-door briefing on the 18th to explain the situation, with outside attention on whether it could impact Taiwan suppliers' supply chain. National Science and Technology Council Minister Wu Cheng-wen today attended the NARLabs National Center for High-performance Computing achievement presentation "From R&D to Industry: Building Taiwan's AI Innovation Ecosystem," and said in an interview that global technology firms all see the continued development trend of AI, and he does not believe a single incident will affect human society's demand for artificial intelligence.
Stock Volatility Is Short-Term, Taiwan Makers' Business Unaffected
Wu Cheng-wen said recent stock market fluctuations are only a short-term phenomenon, and the government's confidence in AI industry development has not changed. Next year's total government budget has already earmarked related funding, and it will continue to promote Taiwan's sovereign AI development and launch related application model development, allowing the AI industry to form an ecosystem in Taiwan and drive overall industry growth. Wu Cheng-wen noted that the world highly relies on Taiwan's semiconductors and AI computing equipment, and there are many international trade control regulations, and Taiwan will abide by the international order. In addition, U.S. firms are important customers for Taiwan, with business covering global markets, and as long as relevant control regulations are not violated, Taiwan operators' business is currently unaffected.
U.S. Midterm Elections Have Limited Impact
Regarding the Democratic Party's big win in the U.S. midterm elections, with outside worries that data center construction will face stricter regulation and hit the AI industry, Wu Cheng-wen said that whether Republican or Democratic, cooperation between both parties and Taiwan has remained consistent, and there are related regulations under the Republicans as well, and he believes the U.S. midterm election results will have limited impact on Taiwan's industrial development. The NARLabs National Center for High-performance Computing today joined with Academia Sinica to showcase a sovereign multimodal artificial intelligence model, integrating image recognition and behavior analysis technologies, which can be applied in areas such as medical care.
Sovereign Multimodal Model Unveiled, 8-Year NT$100 Billion Plan to Aid Transformation
Wu Cheng-wen said in his remarks that in addition to continuing to optimize the trusted AI dialogue engine TAIDE sovereign AI model, R&D results for a new-generation multimodal sovereign AI model were also announced this time, with Academia Sinica investing in AI visual recognition and multimodal technology R&D, including YOLO object detection and recognition technology, and the further developed General Behavior Analysis model GBA. Related technologies can identify human behavior such as falls through image analysis, combined with speech recognition models for smart notification. Wu Cheng-wen noted that for Taiwan to develop sovereign AI, it needs to build its own application models, and model development relies on computing power. Currently, government-invested computing power is mainly built and operated by the National Center for High-performance Computing, supporting academic research and industrial technology research, while the Ministry of Digital Affairs supports industry, and companies are also investing in various model R&D. As AI applications become more widespread, future computing demand will continue to grow, and the government will accelerate AI computing buildout, investing in computing power needed for R&D while also inviting private investment to expand application-side computing resources. He said computing buildout cannot rely on government alone and also needs private participation, driving economic development through commercial operation and giving operators opportunities to expand into global markets, hoping that academic R&D results can be taken over by industry to further develop commercial value. He believes Taiwan may not excel at developing large foundation models like the United States and China, but can use local data to optimize foundation models and train application models meeting local needs. The government will invest an additional NT$100 billion over the next 8 years to help small and micro enterprises upgrade and transform, hoping to use technological strength to help traditional manufacturing and services accelerate AI adoption, making all industries part of Taiwan's technology industry ecosystem. He stressed that Taiwan should not only develop semiconductors and computing equipment, but also develop various innovative application systems in the future to compete for international business opportunities.