DeepMind Chief Meets Heads of Korean Conglomerates

According to MK News, Demis Hassabis, co-founder and CEO of Google DeepMind, visited Seoul and met with President Lee Jae Myung at Cheong Wa Dae on April 27. MK reports that Hassabis held separate meetings with the heads of major Korean conglomerates, including Samsung Electronics Chairman Lee Jae-yong, SK Group Chairman Chey Tae-won, Hyundai Motor Group Chairman Chung Eui-sun, and LG Chairman Koo Kwang-mo. MK reports that Samsung DX head Roh Tae-moon and senior executives from SK hynix and SK Telecom attended the respective meetings.

MK reports that discussions were expected to cover cooperation on AI and semiconductor supply for AI infrastructure. MK's coverage notes that Samsung supplies memory semiconductors and high-bandwidth memory (HBM) to Google, and that SK hynix supplies high-performance HBM for Google TPUs. MK also reports that Google DeepMind develops Google s AI model Gemini, and that Gemini has been embedded in Samsung Galaxy devices per MK's article. The Korea Times reports that Google is planning an AI campus in Korea, which local reporting frames as related context for the visits.

Industry context: Companies training and deploying large foundation models increasingly consume large volumes of memory and HBM, and major cloud and model developers commonly engage semiconductor suppliers and local partners to secure capacity and integration paths. Industry-pattern observations: Korean conglomerates that combine foundry, memory, and consumer-device businesses are frequent partners for cloud and AI platform vendors because they can supply both chips and device-level integrations.

Editorial analysis: The reported meetings bring together a major model developer, national government, and vertically integrated chip and device makers at a strategic moment for compute and memory demand. Industry-pattern observations: When model developers and hardware suppliers hold executive-level talks, observers often see accelerated MOUs, pilot projects, or supply agreements emerge later, especially around HBM capacity and optimized memory architectures.

Observers should watch for follow-on announcements, such as memoranda of understanding, capacity-expansion commitments from memory or foundry units, formal partnership statements, or a public timeline for the Korea AI campus reported by The Korea Times. Also watch product-level integrations of Gemini in consumer devices and any disclosures about HBM supply for TPU-class accelerators.