'Fruits of Supercomputer Adoption' Naver Takes a Step Closer to Securing AI Sovereignty

One Year Since Supercomputer Introduction... Focus on AI R&D
'HyperCLOVA' Large-Scale AI with Growing Versatility
Total of 58 Regular Papers Presented at Top-Tier AI Conferences... Also Prominent at Global Conferences

[Asia Economy Reporter Kang Nahum] Naver, which introduced a supercomputer about a year ago under the banner of securing "Artificial Intelligence (AI) sovereignty," is recently reaping the fruits of success with consecutive commercializations of its services. It is now evaluated to be on par with global big tech companies that have invested massive capital and competed in developing supercomputers themselves.

◆ Nearly 1 Year Since Supercomputer Construction = According to Naver on the 10th, the company built a 700-petaflop (one quadrillion calculations per second) class supercomputer in October last year to strengthen its AI technology competitiveness, establishing infrastructure capable of processing large-scale data. It is the first among domestic companies.

This move is interpreted as Naver actively joining the expanding trend worldwide where big tech companies are utilizing supercomputers to pioneer the cloud business market integrated with new technologies. Especially, Naver's approach focuses on AI R&D. Ultra-large-scale AI requires learning from vast amounts of data, and infrastructure is essential for that. Based on the supercomputer, Naver developed its own ultra-large-scale AI called ‘HyperCLOVA.’

Approaching the first anniversary of the supercomputer's construction, services utilizing HyperCLOVA are also bearing commercial fruits. Naver upgraded the voice recognition engine ‘NEST’ based on HyperCLOVA and applied it to Clova Note to improve recognition accuracy. HyperCLOVA technology was also used in the ‘AI Review Summary’ feature, which analyzes numerous product reviews within Smart Store and extracts a single sentence that best represents the product's characteristics. The supercomputer is also being used to develop a Japanese-specialized ultra-large language model capable of natural Japanese expressions.

Naver is also standing out at global academic conferences. To date, Naver has published a total of 58 regular papers at top-tier global AI conferences, surpassing last year's achievement of 43 papers.

◆ "Securing AI Sovereignty" Supercomputer Survival Competition = The reason why global big tech companies including Naver are actively competing to introduce supercomputers is the judgment that building basic infrastructure such as supercomputers is essential to survive in the 4th Industrial Revolution era centered on new ICT technologies like AI and big data. An industry insider said, "Although the introduction timing was somewhat late, Naver's supercomputer introduced this time is top-tier if limited to AI research," adding, "It could rank high in the global Top 500 supercomputer performance rankings."

Google's high-performance computer ‘TPU v4 Pod,’ announced last May, is estimated to have such fast calculation speed that it ranks within the top 5 worldwide, including all supercomputers developed under national initiatives. It is expected to be used in services such as Google Assistant, automatic translation, and search engines.

Microsoft (MS) has started providing its own supercomputer externally on the cloud-based (Azure) platform since June. It supports research by OpenAI, which developed the language AI (GPT-3) that generates natural sentences like humans, as well as OpenAI research within the United States.

NVIDIA has been operating the supercomputer Cambridge-1, designed and built for research purposes, in the UK since July. AstraZeneca, GlaxoSmithKline, and others are using it for brain disease research and new drug design through AI.

Additionally, Tesla, which introduced supercomputers for autonomous vehicle R&D, is supporting technology development by processing vast amounts of data in autonomous driving development equipped with cameras.

Naver's investment in supercomputers particularly emphasizes "securing AI sovereignty." Naver's judgment is that if future essential AI technologies are not researched directly, services will become dependent on global companies. Ha Jung-woo, head of Naver AI Lab, said, "As a leading AI company, we will strive to secure global AI competitiveness through deep, long-term advanced research and to further solidify the domestic AI technology ecosystem."

This content was produced with the assistance of AI translation services.