As global urbanization accelerates and populations significantly age, cities worldwide are investing heavily in smart city development. The goal is to create safer and more efficient urban environments through AI and digital technologies, enhancing citizens' quality of life. With the emergence of Vision-Language Models (VLM) combined with generative AI, smart city applications have entered a new phase - enabling systems to automatically interpret traffic flow, pedestrian behavior, and unexpected incidents from video footage, and thereby supporting real-time decision-making.In his keynote address ahead of COMPUTEX 2025 in Taipei, NVIDIA CEO Jensen Huang highlighted Kaohsiung's "Smart Kaohsiung Lighthouse Project," which leverages a VLM-based digital twin architecture to interconnect tens of thousands of city surveillance cameras, significantly improving the efficiency and service capabilities of the municipal government. The key technology driving this initiative comes from Linker Vision, a company renowned for its expertise in computer vision technology.Joining the NVIDIA ecosystem to advance smart applicationsWith more than a decade of experience in AI-powered image recognition, Linker Vision has become a trusted partner for many leading global brands. Following the rapid rise of generative AI in 2023, the company began applying VLM's powerful visual comprehension and semantic reasoning capabilities to achieve a deeper understanding of real-world scenes.Unlike traditional object recognition models, VLM can identify a broad range of objects without additional training while discerning contextual relationships among them. This enables the system to detect complex incidents and assess damage that conventional models often fail to recognize. Building on its solid foundation in visual recognition, Linker Vision utilizes VLM's multimodal data processing capabilities to develop a suite of solutions for smart city and smart factory applications.Yi-Ting Liu, Senior Sales Manager at Linker Vision, explained that the company's smart factory solutions are designed to address enterprises' dual challenges of improving workplace safety and operational efficiency. In smart safety applications, the system can accurately detect fires, chemical leaks, or safety violations - such as workers entering restricted areas or failing to wear protective gear - and even identify complex behaviors like "smoking within a restricted zone," triggering composite alerts in real time. Beyond safety, the solution also monitors employee productivity, status, and compliance with standard operating procedures (SOPs), providing factory managers with a visualized, data-driven management tool that effectively reduces human error.Building global presence through AWSLinker Vision has built its solutions on Amazon Web Services (AWS) for many years, achieving outstanding results in the smart city and smart factory domains through modular and standardized system architectures. Yi-Hsuan Lin, Product Manager at Linker Vision, noted that the company extensively utilizes AWS's AI computing power and services, including Amazon Elastic Kubernetes Service (Amazon EKS), Amazon Elastic Compute Cloud (Amazon EC2), and Amazon Bedrock, a fully managed service that offers a choice of high-performing foundation models (FMs) and tools to deploy and operate agents.The key advantage, Lin said, lies in AWS's on-demand flexibility, which enables the team to access the latest GPU resources anywhere in the world, substantially reducing hardware procurement and maintenance costs. When conducting proof-of-concept (PoC) trials across borders, deployments can be launched instantly on AWS's local infrastructure closest to customers, eliminating the time and expense required for purchasing and installing physical hardware. Amid the global push toward Sovereign AI, Linker Vision also offers hybrid cloud and on-premises services, helping enterprises optimize AI models, develop tailored solutions, and ensure data privacy compliance.Expanding internationally with AWS JIC supportLiu emphasized that participation in the AWS Joint Innovation Center (JIC) has laid a strong foundation for Linker Vision's market development and international expansion. Beyond providing abundant cloud resources and technical expertise, the JIC offers a comprehensive set of empowerment services, including business networking, corporate matchmaking, and marketing exposure, helping Linker Vision connect with strategic partners and accelerate business growth.With rapid momentum in the Taiwan market, the company now aims to expand globally-targeting the Middle East, Southeast Asia and the United States-to capture the vast opportunities in smart city and smart factory development.

In modern competitive sports, the margin between victory and defeat often depends on who can capture and analyze data more effectively. Taiwan's triumph at the 2024 WBSC Premier 12 was not only the result of exceptional player performance and precise coaching but also the scouting team's ability to provide rapid, on-site opponent analysis. As a key member of the scouting team, StatsInsight played a pivotal role. Leveraging its stadium optical tracking system and StatsInsight analytics platform - both built on Amazon Web Services (AWS) - the company provided real-time data insights to players and coaches, making it one of the unsung heroes behind Taiwan's championship win.StatsInsight co-founder Leh Shyu noted, "As sports technology evolves rapidly, baseball has long moved beyond the era of paper scorecards. Professional leagues in both the U.S. and Taiwan have adopted electronic and camera-based systems to record every detail of practices and games. StatsInsight is the only company in Taiwan focused exclusively on baseball analytics and sports technology integration. Our goal is to build a solid foundation for the next generation of baseball data development in Taiwan by delivering comprehensive solutions and customized services that bring technology into every ballpark."Products validated through real-world performance: adopted by pro and national teamsBaseball is a sport defined by subtle details and strategy that are difficult to capture with the naked eye. Founded in 2020, StatsInsight has long specialized in baseball data collection and analysis. Its "Electronic Strike Zone" system has become a familiar presence to Taiwanese baseball fans, offering advanced statistics derived from ball flight trajectories and player motion reconstruction.StatsInsight's scouting data systems and analytics services have been widely adopted and highly recognized by both professional baseball teams and the national team.A former baseball player himself, Shyu entered the Chinese Professional Baseball League (CPBL) draft in 2019. Although not selected, his deep knowledge and experience earned him a place on the national team's scouting staff. That opportunity inspired him to co-found StatsInsight with like-minded partners to develop Taiwan's own baseball scouting system.Shyu explained that while the U.S. has a more mature sports technology ecosystem, its systems are often expensive and come with language and support barriers for Taiwanese teams. StatsInsight, though a later entrant, built its system in-house and proved its reliability through multiple competitions. With its ability to deliver customized solutions and localized support, the company successfully partnered with the national team to build an international scouting platform and has since collaborated with professional clubs.Leveraging AWS services to realize a hybrid cloud architectureStatsInsight has deployed its full system and website on Amazon Elastic Compute Cloud (Amazon EC2) and uses Amazon Relational Database Service (Amazon RDS) for managed database operations. Video data is securely stored in Amazon Simple Storage Service (Amazon S3), while Amazon CloudFront Content Delivery Network (CDN) ensures smooth and fast content delivery for optimal user experience. The company is also developing an optical computer vision system and, through resources provided by the Startup Terrace Kaohsiung AWS Joint Innovation Center (JIC), received direct technical guidance from AWS experts to implement and validate its hybrid cloud architecture.Shyu shared, "U.S.-made Hawk-Eye systems are highly advanced and purpose-built for baseball stadiums, but renting one costs around NT$7 million per year, with a minimum four-year lease. Including technical support, total expenses can exceed NT$30 million. Our vision system is designed to match the performance of those systems, combining similar-grade cameras with small on-premises servers and AWS cloud services to deliver a more cost-effective hybrid solution. Taiwan's next baseball era will require comprehensive data and biomechanical analysis. By developing our own system, we aim to provide an affordable and effective option to drive Taiwan's baseball technology transformation."Currently, StatsInsight is collaborating with the National Sports Science Institute in Kaohsiung to deploy its system in baseball stadiums and create a next-generation smart baseball field. This collaboration will also serve as a testbed for validating the company's self-developed optical vision system and as a model for future smart stadiums in Taiwan.

The AI-driven paradigm shift is prompting manufacturers to extend real-time data optimization from management decisions to every detail of the production line. Despite advances in automation, up to 30% of production processes still rely on manual operations. PowerArena, a smart manufacturing solution provider, combines industrial data with AI computer vision to close this "last-mile" gap in automation. Through its platform-based model, PowerArena establishes a mechanism for continuous optimization, helping manufacturers accelerate their digital transformation."Pyramid architecture" enhances factory data utilization and boosts ROI in smart manufacturingAccording to Ian Peng, Solutions Director at PowerArena, the company's core platform, the Human Operation Platform (HOP), integrates AI computer vision and IoT technologies to continuously collect and analyze workstation images and data. It identifies operation cycles, idle times, and SOP compliance, replacing traditional walk-around management or paper-based records. The platform directly addresses three major pain points for manufacturers: increasing human-factor variability caused by labor shortages, management gaps arising from production line relocation, and the lack of real-time visibility in quality control.HOP has been widely adopted in EMS electronics, semiconductor, and automotive assembly line management. Using a pyramid-architecture approach, the platform first collects foundational workstation data through IoT implementation, then applies AI vision to analyze manual assembly processes, and finally focuses on the top 1 to 3% of critical production steps to ensure SOP precision-maximizing economic efficiency and investment returns for manufacturers.PowerArena is also developing a large language model (LLM) specifically for manufacturing applications. The model integrates industry knowledge with platform data to provide real-time production line analysis, data aggregation, and improvement recommendations. Peng noted that this feature is especially valuable for new factory training, as it shortens employee onboarding and enhances productivity.To ensure continuous optimization and rapid scaling of both the HOP platform and the LLM, data infrastructure plays a crucial role. Peng explained that manufacturing data is highly proprietary, difficult to obtain, and massive in scale. To manage this, PowerArena leverages Amazon Web Services (AWS) cloud infrastructure. Using Amazon Simple Storage Service (Amazon S3), it securely stores authorized images and generative AI simulation data for model training and optimization. Once uploaded to the cloud, the data is analyzed and retrained through high-performance computing resources, enabling continuous AI refinement.PowerArena also utilizes Amazon Route 53 for DNS management and AWS Fargate for serverless container deployment in hybrid environments, allowing for flexible scaling, maintenance, and multi-region deployment. The upcoming launch of the AWS Asia Pacific (Taipei) Region in 2025 further strengthens enterprises' confidence in AWS's security and compliance standards. These cloud advantages not only accelerate product iteration and time-to-market but also empower PowerArena with greater deployment agility-serving as a key enabler of innovation.By integrating the HOP platform with AWS cloud services, PowerArena has already delivered tangible results for global manufacturers. A top-five global electronics manufacturer improved production capacity by 5% and reduced yield fluctuation by 70% within just four weeks. Another semiconductor materials company uses AI computer vision to monitor high-risk operations, ensuring employee safety and regulatory compliance while advancing its ESG performance.PowerArena leverages AWS JIC to expand globally and Startup Terrace Kaohsiung to grow market footprintFrom its inception, PowerArena has maintained a strong international focus, with operations spanning North America, Mexico, East Asia, and Southeast Asia - the latter seeing particularly rapid project growth in recent years. Its global strategy centers on replicating best practices: digitizing high-yield production data from factories in Taiwan and China and transferring it to new plants, enabling faster ramp-up and stable capacity contributions for multinational enterprises.In Taiwan, Kaohsiung's industrial landscape is shifting from traditional heavy industries toward high-tech and smart manufacturing. To deepen local engagement, PowerArena established an office at Startup Terrace Kaohsiung through its participation in the AWS Joint Innovation Center (JIC) program. This new base reinforces PowerArena's long-term commitment to the region and serves as a key hub connecting southern Taiwan's industrial ecosystem.Looking ahead, PowerArena will continue leveraging industrial data and AI technologies as its core competencies "accelerating international expansion while strengthening local partnerships" to play a leading role in driving the next wave of smart manufacturing upgrades.