The Endless AI Model Race: Are Models Becoming Commoditized?
Session Overview
Generative AI is evolving beyond simple chatbots into a general-purpose intelligence platform capable of understanding and executing human knowledge and actions.
However, today’s foundation models still face significant limitations, including reliability across long reasoning processes, understanding of the real world, effective agent use, multimodal integration, access to high-quality data, and the cost of training.
Drawing on firsthand experience developing foundation models, this session will examine where AI stands today and explore which technologies are likely to determine the next five years of competition. It will also consider the opportunities Korea may have in the global foundation model race.
Speakers
Ethan Cho is a venture investor and Partner & CIO at TheVentures. He graduated from Seoul National University with a degree in Business Administration and began his career at Samil PwC. He went on to earn his MBA from Columbia University in the United States.
He has since held roles at a global hedge fund, Samsung Electronics' Samsung Strategy & Innovation Center (SSIC), as Korea lead for Qualcomm Ventures, and as a Partnership Development Manager at Google Korea.
After serving as Head of Division at KB Investment and Partner at Fast Ventures, he now leads early-stage startup sourcing and portfolio value-up strategy at TheVentures. Drawing on business development experience at global tech companies including Google, Qualcomm, and Samsung, along with hands-on international investment experience, he excels at reading subtle shifts in founders and markets and quickly turning opportunities into action.
Junghwan Lim is CEO of Motif Technologies, an AI deep-tech company that researches, develops, and delivers generative AI models — including LLMs and image/video generation.
Motif Technologies builds across the full stack of AI technology, from infrastructure and GPU systems to models and services, and Junghwan also serves concurrently as the company's CTO.
Jay Shin is a leading AI researcher in the field of large language models, with more than 4,000 academic citations. He is a co-first author of the Prometheus series, widely regarded as a standard open evaluation model for LLM-as-a-Judge research and published at ICLR 2024. He also received the NAACL 2025 Best Paper Award for BiGGen Bench, an LLM-based evaluation benchmark.
After working at Amazon Alexa AI, he joined Naver and contributed to the development of HyperCLOVA X, Korea’s flagship hyperscale language model. In 2024, he founded Trillion Labs.
Within one year of its founding, Trillion Labs became the first company in Korea to train a 70B-parameter LLM entirely from scratch and release it as open source. The company continues to advance innovative research, including rBridge at ICLR 2026 and gWorld at ICML 2026, described as the world’s first mobile world model.
Today, Trillion Labs is expanding beyond conventional LLMs to develop world models that can understand and operate real-world physical infrastructure. Its proprietary models have been deployed and are currently operating at large-scale power plant sites exceeding 1GW in capacity.