- Large-scale language models (LLMs) demand substantial capital investment and computational resources.
- Korea emerges as a key player in LLM development, following the lead of the United States and China.
- Japan’s strategic moves in AI include subsidizing Softbank’s AI supercomputer development and nudging Naver to divest its stake in Lineyahu.
- The Lineyahu joint venture between Naver and Softbank becomes a focal point amidst Japan’s broader ambitions in AI and Cloud.
- The global market for LLMs reflects a complex interplay of technological advancement, national interests, and strategic partnerships.
Main AI News:
Large-scale language models (LLMs) exceeding 100 billion parameters, akin to the synaptic connections in the human brain, necessitate substantial financial investment. Procuring such models entails not only attracting top-tier talent and fostering their development but also harnessing significant computational resources like data centers. Mere financial prowess doesn’t guarantee success in this arena. The global demand for advancement has created a formidable queue at Nvidia’s doorstep, which commands over 80% of the artificial intelligence (AI) accelerator market.
In the competitive landscape, acquiring an H100 chip, valued at 50 million won, translates to a wait of 50 weeks from order placement to shipment. Consequently, LLM development becomes a testament to a nation’s technological prowess. Korea has notably managed to establish LLM capabilities at a commendable pace, following in the footsteps of the United States and China. Key players in this endeavor include Naver’s HyperClova X, LG AI Research Institute’s Exemployee, and SK Telecom’s AdTX.
Japan’s apparent push for Naver to divest its stake in Lineyahu under the guise of administrative guidance raises questions. This move seems disconnected from mere efforts to secure AI sovereignty. Interestingly, the Japanese government made this announcement concurrently with Softbank’s declaration that Lineyahu would terminate all collaborative ties with Naver.
Simultaneously, Japan’s Ministry of Economy, Trade and Industry emphasized its commitment to allocate 42.1 billion yen (approximately 370.1 billion won) as a subsidy for Softbank’s development of AI supercomputers. Recognizing the necessity of vast chip resources for AI learning, the government aims to procure and provide these resources swiftly. With Japan lacking a significant LLM presence, this initiative indicates the government’s intent to spearhead infrastructure development in this domain.
Lineyahu, a joint venture between Naver and Softbank, serves as a Japanese entity encompassing messaging, search, and payment services. When Naver and Softbank forged this partnership three years ago, Softbank assumed managerial responsibilities while Naver led in technology. Naver harbored aspirations of exporting AI technology to Japan. However, it appears that the Japanese government’s actions extend beyond mere security concerns. This move signals Japan’s broader ambitions, leveraging opportunities in AI and Cloud to bridge gaps across various sectors, positioning itself to compete with Korea.
Conclusion:
Japan’s maneuvers in the AI sector, including subsidies for AI supercomputer development and pressure on joint ventures like Lineyahu, underscore its strategic ambitions. This signals a heightened competition in the global market for large-scale language models, with Korea and Japan vying for leadership positions amidst evolving geopolitical dynamics. Companies and investors in this space must navigate complex regulatory environments and geopolitical tensions while capitalizing on emerging opportunities for innovation and collaboration.
