In a strategic move to counter Nvidia’s dominance, Naver and Intel have teamed up with the Korea Advanced Institute of Science and Technology (KAIST) to forge a new artificial intelligence (AI) semiconductor ecosystem.
The partnership was cemented on April 30 at the KAIST main campus in Daejeon, where the parties established the Naver-Intel-KAIST AI Joint Research Center (NIK AI Research Center). The agreement includes the establishment and operation of the center, focusing on developing AI semiconductor technologies.
Leading the new center is KAIST’s Professor Kim Jeong-ho from the Department of Electrical Engineering and Computer Science, a renowned scholar in AI semiconductor design and AI applications (AI-X). Lee Dong-su, an expert in AI semiconductor design and AI software from Naver Cloud, will co-direct the center.
Approximately 20 faculty experts in AI and software from KAIST, along with over 100 master’s and doctoral students, will participate as researchers at the center. While the center is initially set to operate for three years, this period may be extended depending on the outcomes and the needs of the participating organizations.
This initiative marks the first time that Intel, a global semiconductor giant, has established a joint research center at a Korean university to develop software platforms for AI semiconductors. The center aims to jointly build a platform ecosystem for Intel’s Habana Labs-developed AI training and inference chip, “Gaudi,” over two years, undertaking 20 to 30 industrial-academic research projects.
Naver and Intel will provide the ‘Gaudi2’ based on the Naver Cloud platform to the KAIST joint research center. KAIST researchers will utilize Gaudi2 for their studies, publishing research papers annually. In addition, they will share various infrastructure facilities and equipment necessary for joint research, and continue cooperative activities including personnel exchanges.
Professor Kim Jeong-ho stated, “Through the use of the Gaudi series, KAIST can secure technical know-how in AI development and semiconductor design and operation software development. Particularly, the establishment of this joint research center is highly significant as it enables us to gain experience in operating large-scale AI data centers and to secure the necessary AI computing infrastructure for future research and development.”
Lee Dong-su, CEO of Naver Cloud, expressed his expectations: “Alongside KAIST, Naver Cloud is leading various research initiatives and hopes to expand the AI ecosystem centered around ‘HyperClova X’.”
Earlier this month, Naver and Intel also announced their cooperation in the AI field at Intel Vision 2024 held in Phoenix, Arizona. Previously, AI developers primarily used Nvidia’s CUDA programming platform, but Intel and Naver’s move to establish their own ecosystem marks a significant shift. The sequential partnerships with Naver and now KAIST are viewed as Intel’s high regard for Korea’s AI capabilities.
