The New Paradigm of the Global Tech Industry by the End of 2025: From Metaverse to AI Agents
Editor
From Metaverse to AI Agents: A Major Shift in Tech Investment Trends
As of the end of 2025, the global tech industry is experiencing a rapid transition from metaverse-centric investments over the past three years to generative AI and autonomous agent technologies. According to the latest report by market research firm Gartner, the global AI agent market size is expected to reach approximately $42 billion by 2025, marking a 187% increase from the previous year. Notably, while investments related to the metaverse decreased by 34% compared to 2024, venture investments in AI agents surged by 240% during the same period.
This shift is driven by a change in corporate perception towards practicality and profitability. For instance, Meta (California, USA) recorded a total loss of $58 billion in its metaverse division, Reality Labs, from 2022 to 2024, but by focusing on AI agent development in 2025, it achieved a quarterly revenue increase of over 15%. Microsoft (Washington, USA) also announced that its Copilot agent service generated an additional $2.3 billion in revenue in the third quarter of 2025 alone.
Korean companies are swiftly responding to these global trends. Samsung Electronics (Gyeonggi-do, Korea) began mass production of AI-dedicated semiconductors, HBM4, in the second half of 2025 and is supplying memory solutions optimized for AI agent operations through collaboration with NVIDIA (California, USA). SK Hynix (Gyeonggi-do, Korea) also announced an investment of 12 trillion won in 2025 alone in CXL (Compute Express Link) memory technology to enhance the real-time learning capabilities of AI agents.
The New Dynamics of Platform Competition: The Race to Build Agent Ecosystems
The most intense competition in the AI agent market is in the area of platform ecosystem building. Currently leading the market is OpenAI’s (California, USA) GPT agent platform, although as a private company, direct investment is limited. Instead, Microsoft is aggressively targeting the enterprise market through its Azure AI Agent Service, backed by a $13 billion investment in OpenAI.
Naver (Gyeonggi-do, Korea) launched the HyperCLOVA X Agent Studio in October 2025, establishing a dominant position in the Korean-language AI agent market. Notably, within two months of its launch, more than 100,000 custom agents were created, with the number of daily active agents exceeding 150,000. This is attributed to its Korean language processing capability, which is 23% higher on average compared to global platforms.
Kakao (Jeju-do, Korea) is also attempting a differentiated approach through its AI agent service ‘Kakao i Agent’ on the KakaoTalk platform. Based on the KakaoTalk ecosystem with 50 million daily active users, it provides personalized AI assistant services, with monthly active agent users surpassing 8 million as of the fourth quarter of 2025. This has resulted in a 45% increase in the proportion of AI service revenue in Kakao’s total revenue compared to the previous year.
Internationally, Google’s (Alphabet, California, USA) Gemini agent and Amazon’s (Washington, USA) Alexa agent are demonstrating unique competitiveness in the search and voice interface domains, respectively. Google’s Gemini agent is directly competing with Microsoft in the enterprise productivity tools market through integration with Google Workspace, with 250 million users worldwide utilizing Gemini agent features as of 2025.
In the Chinese market, Baidu’s (Beijing, China) Ernie Bot agent and Alibaba’s (Hangzhou, China) Tongyi Qianwen agent are providing specialized services in the search and e-commerce domains, respectively. Notably, Baidu reported that the daily query count through the Ernie Bot agent platform exceeded 1.5 billion in the third quarter of 2025, with AI services accounting for 32% of its total revenue.
A key point in this platform competition is that each company is attempting to differentiate by combining their existing strengths with AI agents. Search companies are developing agents specialized in information retrieval and analysis, social platforms are focusing on communication and personalization, and cloud companies are creating agents optimized for enterprise workflows. This indicates that the AI agent market is evolving into a competitive landscape across various specialized domains rather than being dominated by a single platform.
From a semiconductor perspective, the hardware requirements to support real-time inference and learning capabilities of AI agents are rapidly increasing. NVIDIA’s H100 and latest H200 GPUs have been in short supply throughout the second half of 2025, acting as a constraint on the infrastructure expansion of AI agent service providers. In this context, Samsung Electronics and SK Hynix’s AI memory technologies are becoming increasingly important, with Korean companies holding over 80% of the global market share in the high-bandwidth memory (HBM) market, which supports long-term memory and learning capabilities of agents.
Another trend emerging alongside the growth of the AI agent market is the integration with edge computing. There is a rapidly growing demand for agents that operate on local devices rather than the cloud to ensure real-time responsiveness and privacy protection. Qualcomm’s (California, USA) Snapdragon 8 Gen 4 processor is equipped with an NPU (Neural Processing Unit) specialized for on-device AI agent operation, enabling high-performance agent services on smartphones and tablets. This is becoming a new growth driver for the mobile AI agent market, with approximately 35% of total AI agent processing being performed on edge devices as of the fourth quarter of 2025.
Adoption of AI Agents by Industry and Market Outlook
The practical value of AI agent technology is being demonstrated through specific use cases across various industries. In the financial services sector, AI agents are automating tasks such as customer consultations, investment advice, and risk management, resulting in significant cost savings. JPMorgan Chase (New York, USA) announced that it saved $2.8 billion annually in customer service operating costs through the adoption of AI agents in 2025, with customer satisfaction improving by 15%.
In Korea’s financial sector, the adoption of AI agents is also accelerating. Major insurance companies like Samsung Life and Hanwha Life have begun using AI agents for insurance claim processing and underwriting tasks, achieving a 75% reduction in processing time compared to traditional methods. Notably, hybrid models where AI agents collaborate with human experts in complex insurance product design and risk analysis are gaining attention.
In the manufacturing sector, AI agents are becoming a key component of smart factories. Hyundai Motor (Seoul, Korea) introduced an AI agent-based quality control system at its Ulsan plant in 2025, reducing defect rates by 42% and improving production efficiency by 18%. This system analyzes production line data in real-time and predicts potential issues to prevent production stoppages.
Global manufacturers are reporting similar results. Siemens (Munich, Germany) improved equipment utilization from 93% to 97% through a predictive maintenance system using AI agents, saving 1.5 billion euros annually in maintenance costs. Toyota (Aichi, Japan) also optimized its supply chain through an AI agent-based system, reducing inventory costs by 22% while increasing on-time delivery rates to 98.5%.
In the healthcare sector, the use of AI agents is bringing about revolutionary changes, particularly in diagnostic support and treatment planning. Mayo Clinic (Minnesota, USA) introduced a personalized treatment planning system using AI agents in 2025, improving the five-year survival rate of cancer patients by 12%. This system analyzes genetic information, medical history, and lifestyle patterns to recommend optimal treatment methods.
In Korea, Seoul National University Hospital and Samsung Medical Center have implemented AI agent-based diagnostic support systems, resulting in an average 16% improvement in diagnostic accuracy by medical staff. Notably, in the field of imaging diagnostics, AI agents have collaborated with radiologists to achieve a 23% increase in early cancer detection rates.
In the retail and e-commerce sectors, AI agents are being used to provide personalized shopping experiences and optimize inventory management. Amazon (Washington, USA) announced that its AI agent-based recommendation system improved customer conversion rates by 35%, equivalent to an annual revenue increase of $18 billion. Additionally, it improved picking efficiency by 28% and reduced delivery times through the use of AI agents in warehouse management.
Domestically, companies like Coupang and 11st are enhancing customer service and product recommendation systems using AI agents. Notably, Coupang’s AI agent-based chatbot reduced customer inquiry response times from an average of 4 minutes to under 30 seconds, resulting in increased customer satisfaction.
In the education sector, the use of AI agents is showing significant results in personalized learning and real-time feedback. Khan Academy (California, USA) reported an average 34% improvement in learner achievement through its AI agent-based personal tutor system. This system analyzes learners’ understanding levels and learning patterns in real-time to propose optimal learning paths and provide additional explanations or practice problems as needed.
In Korea’s education sector, the adoption of AI agents is also expanding. Major educational companies like Daekyo and Woongjin ThinkBig have launched AI agent-based learning platforms to provide customized educational services tailored to individual learners’ levels. Notably, in English and math learning, AI agents analyze learners’ incorrect answers in real-time and suggest improvement strategies, significantly enhancing learning effectiveness.
When considering these industry-specific adoption cases, it is evident that the growth potential of the AI agent market extends beyond a mere technology trend to actual economic value creation. According to the latest research by McKinsey, the economic value that AI agent technology can generate across all industries is expected to reach $2.6 trillion annually, with approximately 40% of this value projected to be realized between 2025 and 2027.
However, alongside this growth outlook, there are risk factors to consider. First, the issues of trustworthiness and explainability of AI agents remain unresolved. In high-risk fields such as healthcare and finance, there is a growing need for systems that allow humans to understand and verify the decision-making processes of AI agents. Second, privacy protection and data security issues are major barriers to the adoption of AI agents. With regulatory tightening, such as the EU’s AI Act, the cost of adopting AI agents for companies is increasing, making market entry particularly challenging for small and medium-sized enterprises.
Third, concerns about job displacement due to AI agents are emerging as a social issue. According to a report by the International Labour Organization (ILO), approximately 85 million jobs worldwide are expected to be affected by AI agents between 2025 and 2030, but simultaneously, 97 million new types of jobs are projected to be created. This underscores the importance of retraining and skill upgrading for the existing workforce.
As of the end of 2025, the AI agent market is transitioning from an initial growth phase to a full-scale commercialization phase. In the next 2-3 years, companies that succeed in this market are expected to be those with not only technological excellence but also the ability to create tangible business value and respond to regulations. Particularly for Korean companies, there is an opportunity to play a significant role in the global AI agent ecosystem, leveraging their strengths in semiconductor and platform technologies.
The information included in this analysis is based on publicly available data and industry reports, and additional due diligence and expert consultation are recommended when making investment decisions.
