"Next Market Leader After Semiconductors" Agentic AI

This suggests that AI is evolving beyond simply generating information to acting in the real physical world.

Agentic AI goes beyond simply answering human questions; when given a goal, it autonomously formulates a plan, goes through multi-step task processes, and even reviews the final results. This means it can act as a secretary with autonomous problem-solving capabilities, overcoming the limitations of existing AI.

This shows the potential for AI to become a key tool for improving productivity by deeply engaging in actual work processes beyond information retrieval.

OpenInterpreter is an open-source project started by one developer, allowing AI to autonomously read and write files, browse the web, and perform various tasks that a human can do. This presents the possibility of a personalized AI assistant and shows that an era where AI can autonomously handle users' complex requirements is approaching.

This case demonstrates that Agentic AI can provide revolutionary efficiency even in complex and specialized fields.

Jensen Huang emphasized that AI has entered a stage where it drives economic growth, rather than just being a cost-saving tool. He notably coined the phrase 'computing power is revenue,' stressing that in the age of AI, superior computing power will directly lead to corporate profit generation. This suggests that investors should consider the computing power of AI-related companies as an important investment indicator.

In AI investment, it is crucial to judge based on the overall market direction rather than short-term growth figures that change every quarter. Currently, the most active area for Agentic AI is coding and programming, which is an important indicator of the practical applicability of AI technology.

Agentic AI is currently exerting a powerful influence in the coding sector, but its expansion will not be limited to coding. It is expected to spread to all knowledge-based tasks such as document creation, data organization, and report writing, as well as to design and quality control in industrial sites, and even to personal daily tools. Therefore, evaluating the potential market for Agentic AI by limiting it to specific figures would be an underestimation of its broad scalability.

Here, a harness signifies infrastructure, similar to the tackle a horse needs to pull a carriage. He emphasized that AI models alone cannot perform actual tasks, and physical and software infrastructure to operate and support them is essential.

Just as a highly intelligent new employee cannot produce results without a desktop, network connection, database access rights, a work reporting system, and security regulations, the adoption of Agentic AI by companies involves building both model-based and physical/managerial foundations in an integrated manner.

Given that AI operates 24/7, a stable cloud infrastructure is an essential component, so the ETF adopts a strategy of concentrating investments in companies holding core technologies of the Agentic AI ecosystem.

This ETF was designed with Anthropic and OpenAI, currently privately held, in mind as TOP2 candidates for the future. Rather than pre-selecting specific companies, it leaves open the possibility of these companies being included in the ETF if they go public, according to the selection criteria. This reflects a flexible strategy to respond to the growth of key AI companies that are currently privately held.

It also includes essential components for AI operations such as cloud infrastructure, security, and data management companies in its portfolio, covering companies that contribute to the overall ecosystem's growth.

This ETF comprehensively includes the necessary elements for implementing capable Agentic AI. It comprises a comprehensive investment portfolio including essential underlying infrastructure companies for agent operation, such as Dell (computer and desktop infrastructure), Arista Networks (connectivity infrastructure), Snowflake (data management), Datadog (monitoring), Palo Alto Networks, and CrowdStrike (security solutions).

Department Head Kim Seung-cheol suggested allocating 60% of surplus funds to themes expected to see structural growth with AI development. Specifically, he proposed distributing to semiconductors (25%), power and nuclear energy (15%), and early themes like Agentic AI (20%), while holding the remaining 40% in cash and short-term bonds to prepare for market corrections and utilizing opportunities for phased buying during such corrections, presenting an aggressive portfolio strategy.

AI investment is now entering the third stage, 'putting AI to work,' beyond the first (preparation) and second (model development) stages. This signifies a crucial turning point where AI transforms from a mere tool into a workforce utilized across corporate settings and daily life. The HANARO US Agentic AI TOP2+ ETF was developed with this shift in AI investment stages and the rise of Agentic AI as its core investment points.