South Korea’s Samsung Electronics (005930.KS) is accelerating its “agentic shift” strategy to embed artificial intelligence into real-world work environments and organizational culture. The vision goes beyond AI model performance competition to infuse AI across products, manufacturing, design, and workflows — converting intelligence into measurable impact.

Samsung Electronics hosted the “Samsung AI Forum 2026” on September 30 at its Seocho headquarters in Seoul’s Seocho District. Marking its 10th anniversary, this year’s theme was “The Agentic Shift: From Intelligence to Impact.” Around 400 attendees gathered on-site, including 100 professors and researchers from South Korean AI graduate programs and 300 Samsung Electronics employees in AI-related roles, with the event also livestreamed on YouTube.

In his welcome address, Jun Young-hyun, Vice Chairman and CEO of Samsung Electronics, identified “harmony” as the new challenge posed by AI’s proliferation. “AI is a core engine driving innovation in our daily lives, while simultaneously presenting new challenges around security, trust, and harmony with the business environment,” he said.

He added, “Agentic AI will bring about fundamental changes in how we work and in organizational culture, going beyond mere technological innovation. We will seek direction together with global AI leaders and drive AI technology innovation.” The message underscored that the focus should be on the societal ripple effects when multiple technologies converge, rather than on technological advancement alone.

Physical AI and Continual Learning Emerge as Core Technologies

At this year’s forum, Samsung Electronics put physical AI and continual learning front and center as the key technologies to realize the agentic shift.

In the physical AI domain, the company unveiled research on “world models” — systems that learn how the real world operates to predict future scenarios and simulate the outcomes of actions — as well as robot-based AI models. The continual learning session introduced “Personalizing AI,” which learns new information while retaining existing knowledge, and “Self-Evolving AI,” which autonomously analyzes real-world usage patterns and pain points after product launch to improve performance.

AI applications are expanding beyond finished products into semiconductor operations. The Device eXperience (DX) division shared product application cases for on-device AI, physical AI, and continual learning technologies. The Device Solutions (DS) division presented innovation cases applying AI across design, manufacturing, and other workflows.

Afternoon sessions were split between DX and DS divisions. The DX AI Technology track covered efficient AI design principles, optimization techniques, and on-device AI platform applications. The DS AX (AI Transformation) Innovation track shared AI applications already deployed in the semiconductor business, along with AI transformation cases from global partners. Partners including Dell Technologies and Anthropic introduced approaches for deploying trustworthy AI agents in business settings and methods for building enterprise AI factories.

Global Big Tech and Academic Experts Convene

The morning keynote sessions featured leading experts from across the global AI ecosystem.

Richard Ho, Vice President of Hardware at OpenAI, explained how agentic AI — systems that make autonomous decisions based on large language models (LLMs) — will transform enterprises and industries at large. His core message centered on productivity innovation through AI models. A former key member of Google’s early TPU AI accelerator development team, Ho currently leads the “Jalapeño” project, OpenAI’s first AI inference chip co-developed with Broadcom and announced in June.

David Green, AI technology leader for Asia-Pacific and Japan at Amazon Web Services (AWS), presented security and governance strategies for implementing agentic AI safely and cost-effectively in enterprise environments.

From academia, Professor Ranjay Krishna of the University of Washington unveiled a new paradigm for overcoming the reasoning limitations of vision-language models and achieving high-performance robotics AI even with small-scale models. Kim Yoon-hyung, Executive Vice President at Samsung Research (and adjunct professor at MIT), a researcher in generative AI and language models, introduced architectural innovation approaches for next-generation “Agentic Intelligence.”

Samsung Electronics’ decision to foreground agentic AI’s organizational culture impact at the 10th edition of its AI forum is widely interpreted as a dual strategy: securing technology leadership while simultaneously strengthening internal innovation momentum. As assessments mount that AI model performance competition is reaching its limits, the move signals a strategic pivot toward real-world application and value creation.

Notably, the disclosure of AI applications in semiconductor design and manufacturing highlights Samsung Electronics’ unique dual position as both a consumer and supplier of AI technology. The strategy is seen as a compound approach: applying AI to its own semiconductor processes to improve yield and efficiency, while leveraging that experience to strengthen competitiveness in the AI semiconductor market.