[Drifting Korean-style AI]③ Even with HyperCLOVA X ecosystem built, investment lags behind... "No killer model"

65 MOUs Related to HyperCLOVA X... Education, Finance, SW, etc.
"Hard to Find Anything That Surpasses Competitors"

Image source=Yonhap News

Developing artificial intelligence (AI) requires astronomical costs. Since continuous learning affects AI performance, funding is a core element of competitiveness. It is difficult for Naver, a leading domestic AI company, to escape this influence. Although Naver collaborates with various companies and institutions through AI models such as the large language model (LLM) HyperCLOVA X, it is evaluated that the company is insufficient in securing development costs.

According to Naver on the 20th, since February last year until now, a total of 65 memorandums of understanding (MOUs) have been signed utilizing HyperCLOVA X. It is estimated that about 2,000 companies and institutions are using Naver’s AI development tool, CLOVA Studio.

The fields of MOUs include public education, cloud, education, gaming, and finance. Naver has joined hands with the Gyeongbuk Provincial Office of Education to innovate future-oriented education and administrative services based on HyperCLOVA X. By applying HyperCLOVA X to the School Support Comprehensive Data Room, it can autonomously recognize and learn vast amounts of data and provide users with related source links that can be directly verified, enabling more accurate and prompt information delivery, which is expected to reduce the workload for school staff.

In the distribution and entertainment sectors, Naver is collaborating with Hyundai Department Store. Hyundai Department Store’s AI Louis, which optimizes marketing writing, uses HyperCLOVA X as its core engine. It intensively learns the sentiment and style of advertising copy and can adjust tone considering the age group. In particular, the first draft of copywriting, which usually took more than two weeks, can be shortened to an average of less than three hours.

Additionally, Naver is partnering with generative AI startup FortyTwoMaru to advance HyperCLOVA X and prepare for the introduction of customized services. It has also been consecutively selected for ultra-large AI-related support projects led by the Ministry of Science and ICT, the National Information Society Agency (NIA), and the National IT Industry Promotion Agency (NIPA).

Naver is increasing external touchpoints for HyperCLOVA X to establish AI monetization. The scale of investment lags behind global standards, and there are many voices that the current possibility of generating profits through investment is low. According to the 'Stanford AI Index 2024' report by the Human-Centered AI Institute (HAI) at Stanford University in the United States, Korea’s private AI investment last year was $1.39 billion (approximately 1.8828 trillion KRW), ranking ninth in the world. The United States led with $67.22 billion (approximately 91 trillion KRW), followed by China ($7.76 billion), the United Kingdom ($3.78 billion), and Germany ($1.91 billion). Considering that financial power determines AI competitiveness, it seems difficult for domestic IT companies including Naver to compete with overseas big tech companies.

Through the advancement of HyperCLOVA X, Naver is enhancing capabilities in writing, reasoning, and planning, while also providing enterprise solutions that strengthen security by linking with internal networks to comply with corporate security policies. In the future, Naver plans to sequentially release HyperCLOVA X models suitable for corporate customers’ usage purposes, including multimodal technologies (simultaneous processing of image, video, audio, and other data).

However, there are still voices expressing disappointment when compared to ultra-large AI models from other global companies. An AI startup official said, "There are many models that can replace HyperCLOVA X," adding, "Although it has excellent performance, it is still difficult to find something that surpasses competitors."

Professor Cho Sung-bae of Yonsei University’s Department of Computer Science said, "Building an LLM requires a lot of investment, and Naver initially showed a strong push forward," but added, "It feels like they are hesitating because continuous updates require enormous funds."