logo

Developers’ View on the Future: Korea’s Largest Developer Conference, “DEVIEW 2023,” Kicks Off
NAVER Cloud Unveils “HyperCLOVA X” to Enable Universal Access to Hyperscale AI and Create a Globally Competitive Hyperscale AI Ecosystem 

- Access for all in the creation of specialized hyperscale AI products for better productivity by incorporating own data with scalability from company to country level
- “The commercialization of Hyperscale AI has been made possible due to proactive investment in infrastructure and the development of native technologies,” says CEO Kim Yoo-won
- NAVER's Hyperscale IDCs and AI semiconductor solutions have facilitated the provision of cost-effective and high-quality AI services, which are anticipated to be the cornerstone of Korea's industrial competitiveness 

 

February 27, 2023

 

The largest developer conference, “DEVIEW 2023,” held by NAVER Corporation (CEO Choi Soo-yeon), kicked off on February 27. Here, NAVER shared its cutting-edge technologies designed for various industries as products of the company’s R&D toward global market expansion. The conference allowed NAVER to emphasize its willingness to communicate with the conference’s participants for the technologies’ advancement. 

 

DEVIEW began with a keynote speech by NAVER Cloud CEO Kim Yoo-won. He highlighted how AI like ”ChatGPT” portends a major change to the world and that by leveraging the synergy between hyperscale AI technology and cloud and SaaS businesses, NAVER Cloud will rise to become a hyperscale tech company leading AI transformation.

[Photo] On February 27, DEVIEW 2023 kicked off with a keynote speech by NAVER Cloud CEO Kim Yoo-won.

 

Kim announced that “HyperCLOVA X,” a hyperscale AI platform that can cater to user needs, will be launched in July. While hyperscale AI technologies such as generative AI are rapidly changing the technology paradigm, Team NAVER, representing Korea’s IT industry, also aims to take the lead in commercializing hyperscale AI to make it accessible to anyone with its plans to create a globally competitive hyperscale AI ecosystem in Korea.

 

According to NAVER Cloud CEO Kim Yoo-won, “NAVER Cloud will become a hyperscale tech company leading AI transformation.”

Team NAVER, equipped for hyperscale AI with consistent and early-stage investments, plans to launch “HyperCLOVA X” in July.

 

Kim expressed confidence that “Team NAVER is well prepared to join the changing trend by concentrating on global-level AI technology and capabilities. HyperCLOVA X is a type of hyperscale AI that combines user data and HyperCLOVA to deliver prompt responses that meet the users’ needs. We expect individual services, businesses, and even countries to develop AI products that cater to their needs and create new business opportunities for greater value.”

 

[Photo] NAVER Cloud CEO Kim Yoo-won talking about HyperCLOVA X at DEVIEW 2023 at COEX, Samseong-dong

 

Since the arrival of ChatGPT, interest in hyperscale AI services has been rising. However, difficulties related to data’s linguistic limitations, reliability, and up-to-dateness, as well as cost issues, still make commercialization challenging. This is why heavy investment in hyperscale AI and its technologies and data is essential to secure the understanding of a specific language and culture and suitability for the industries and the environment of the country. Having already invested in its own infrastructure and technologies, NAVER plans to enhance the industrial competitiveness of Korea’s hyperscale AI industry with partners.

 

After Kim’s presentation, keynote speeches were continued by NAVER Cloud hyperscale AI technical lead Sung Nako, who discussed the competitiveness of “HyperCLOVA X,” NAVER Cloud CTO Kwak Yong-jae, who discussed the essential infrastructure for hyperscale AI, and NAVER Search US chief scientist Kim Yong-bum, who discussed the “‘SearchGPT” project.

 

“HyperCLOVA X” allows the creation of customized hyperscale AI products, boosting productivity even with only small amounts of data.

 

Sung Nako explained the competitiveness and strength of HyperCLOVA X, which is characterized by its functionality and stability, highlighting how “for more professional and advanced hyperscale AI services, learning from data closely connected to the users themselves, such as individuals or companies, is required.

 

Sung also said, “As HyperCLOVA X learned Korean language data 6,500 times more than ChatGPT and has been enhanced to realize the AI desired by users, even small amounts of data can be combined with user data to establish a hyperscale AI product optimized for a specific service or company. It has been upgraded to promptly provide a variety of interfaces that can cater to user needs.” 

 

[Photo] NAVER Cloud hyperscale AI technical lead Sung Nako delivering a keynote speech at DEVIEW 2023

 

NAVER’s hyperscale AI will become stronger with AI semiconductor solutions and hyperscale IDCs.

 

Kwak Yong-jae introduced AI semiconductor solutions and hyperscale IDCs necessary for hyperscale AI services, stressing how “Considering that hyperscale AI will be further advanced in the future, an integrated approach to the software–hardware operating environment is needed.”

 

Kwak added, “We are developing a lightweight AI semiconductor solution with Samsung Electronics that has all the necessary functions for the computation, learning, and deduction of large language models (LLMs) while having a model size of one-tenth the existing GPUs and more than four times the power efficiency. Gak Sejong, the hyperscale IDC which will be opened this year, is equipped with stable power capacity and cooling technology to provide AI infrastructure efficiently. If the AI semiconductor solution is operated at Gak Sejong, it will have four times the operating scale compared to existing GPUs, enabling more efficient and stable operation.”

 

[Photo] NAVER Cloud CTO Kwak Yong-jae during his keynote speech at DEVIEW 2023

 

HyperCLOVA-based hyperscale search technology project “SearchGPT,” deemed trustworthy, connected, and multimodal, was also introduced.

SearchGPT is expected to provide diverse search experiences through the synergy of searching, hyperscale AI, and service.

 

Kim Yong-bum, the last person to deliver a keynote speech, explained NAVER’s next-generation search technology project, “SearchGPT.” The “SearchGPT” project uses LLM OCEAN as its backbone, which optimizes HyperCLOVA for NAVER Search. Based on this, it uses the search flow data accumulated in NAVER for the last 20 years to guide users along the optimal path to achieve their search goals, better understanding their search intent and generating trustworthy answers.

 

[Photo] NAVER Search US chief scientist Kim Yong-bum talking about the “SearchGPT” project at DEVIEW 2023

 

According to Kim, “The project focuses on the three characteristics of trustworthy data, connectedness with NAVER’s services, and multimodal service for efficient delivery of information.” He finishes his speech by emphasizing how “By generating accurate and trustworthy information through up-to-dateness, repeated cross-validation, user feedback, and integration with NAVER’s services, such as NAVER Shopping, Pay, and Map, we plan to deliver information optimized for the user search intent in intuitive formats that utilize images, videos, and audio. We expect to be able to provide a more unique search experience with the Search GPT project.”

 

DEVIEW started as NAVER’s internal technology event in 2008 and opened its doors to external developers in 2010. It has established itself as a representative developer conference where the latest technologies and trends in the IT sector are shared, attracting the attention of developers in various fields in Korea and abroad. The 15th DEVIEW held this year goes on for two days at COEX, Samseong-dong, from February 27 to 28, with 3,500 participants. This conference comprises 46 sessions on various topics, including artificial intelligence and machine learning (AI/ML), cloud, web, search, mobile, natural language processing (NLP), data, and infrastructure. 

 

-END-