Choi Seung-woo and Naver's AI Strategy

Choi Seung-woo and Naver’s Strategic AI Leadership: Translation, Content Generation, and the Future of Sovereign AI in South Korea

Introduction

In recent years, South Korea’s leading internet platform, Naver Corporation, has emerged as a prominent force in the global artificial intelligence (AI) ecosystem. Central to Naver's AI ascendancy is the leadership of Choi Seung-woo, Director of AI Research. Choi has been a driving voice-especially in recent interviews-articulating both the company’s technological ambitions and South Korea’s sovereign approach to AI in an increasingly competitive and geopolitically charged landscape. This report explores key elements from recent public interviews and presentations, unpacks Choi’s role and contributions within Naver, analyzes Naver’s advanced AI translation and content generation platforms, and examines the company's strategic differentiation from global titans like Google and OpenAI.

Building on a diverse array of authoritative sources, the following sections will provide a comprehensive, paragraph-focused analysis of Naver’s technological, organizational, and policy direction. It will cover the design and impact of multilingual models such as HyperCLOVA X, highlight the integration and performance of these models compared to international competitors, and assess broader issues-including Naver’s pioneering work in digitalization of Korean language resources, sovereign AI initiatives, partnerships, and AI ethics. By synthesizing insights from corporate statements, technical reports, news coverage, and recent benchmark results, this report offers a holistic understanding of Naver's multifaceted AI strategy and Choi Seung-woo’s influence within it.

Choi Seung-woo: Recent Interview and Vision for AI in South Korea

The most recent high-profile interview with Choi Seung-woo, Director of AI Research at Naver, shines a spotlight on the pivotal shifts currently underway in South Korea's AI landscape. In this discussion, Choi outlines not only the technical advancements but also the strategic vision underpinning Naver’s pursuit of sovereign AI-an approach tailored to local culture, language, and governance. Choi begins by acknowledging South Korea's impressive technological infrastructure, citing the country’s sixth-place global ranking in AI capacity, yet tempers this with an honest assessment that commercialization trails at eighteenth worldwide. This gap, Choi argues, is both a challenge and an opportunity1.

The interview highlights several key themes: the development and open-sourcing of Naver's flagship large language model HyperCLOVA X, the company’s emphasis on high-context Korean language understanding, and a broader call for responsible, culturally aware AI development. Choi repeatedly contrasts Naver’s strategy with the more generalized, English-centric approaches of Google and OpenAI, emphasizing Naver's focus on “localization.” He notes that this is not simply a matter of language translation, but embedding AI within the nuanced domain of Korean culture, regulatory needs, and economic priorities. Furthermore, Choi underscores the need for a diverse ecosystem of language models-large and small-catering to different verticals and use cases, reflecting a philosophy of ecosystem growth rather than one-size-fits-all dominance1.

Crucially, the interview foregrounds AI safety, ethics, and sovereignty. Choi directly addresses global debates about AI governance, arguing that Korean AI must respect regional diversity, avoid imported biases, and balance rapid innovation with responsible deployment. This position is not only technologically strategic but also politically resonant, aligning with the government’s broader investment in sovereign AI and positioning Naver as both national flagship and international innovator.

Choi Seung-woo's Role as Director of AI Research

Choi Seung-woo has been instrumental in anchoring Naver's AI R&D trajectory at a time when the company seeks both domestic consolidation and global competitiveness. Official directories and professional listings place Choi as a senior research engineer and Director of AI Research at Naver since 2022, after previous research roles at Kakao and other influential South Korean tech firms2. Choi is renowned within the company and South Korea’s research community for bridging advanced machine learning methodologies with practical, large-scale applications tailored to the nation's needs.

In his capacity as Director, Choi leads multidisciplinary teams responsible for the design and improvement of Naver’s large language models, multimodal AI systems, translation pipelines, and content generation tools. His leadership extends to managing collaborations with academic institutions, steering participation in government-led sovereign AI projects, and fostering open-source initiatives to accelerate the local AI ecosystem. Notably, Choi's ability to articulate the social implications and ethical imperatives of AI, as well as technical challenges, positions him as a vital communicator within and beyond Naver.

Choi’s influence is also evident in Naver’s partnerships, ranging from co-development with leading universities such as KAIST and Seoul National University to cross-industry projects with cloud infrastructure providers and international technology firms. This collaborative focus reflects Choi’s broader understanding that innovation in AI-especially for a language and culture as unique as Korean-requires cross-sectoral momentum and community-building beyond the company’s walls.

Research Contributions of Choi Seung-woo

Beyond his organizational leadership, Choi Seung-woo boasts a track record of research that bridges theory and practice within natural language processing (NLP), human-computer interaction, and cross-lingual modeling. Notable recent contributions include the development of advanced techniques for machine translation and information retrieval, such as “BridG MT: Enhancing LLMs' Machine Translation Capabilities with Sentence Bridging and Gradual MT”, published at the 63rd Annual Meeting of the Association for Computational Linguistics in July 20253. This work explores novel methods to improve large language models’ translation quality, particularly for complex sentence structures-directly informing the design of Naver’s Papago and HyperCLOVA-based translation services.

Choi’s earlier work focused on multimodal content analysis, topic category extraction across media, and user interaction studies for video and sports content-all highly pertinent to Naver’s core services such as Webtoon, Naver TV, and its live streaming ecosystem4. His research addresses both theoretical performance metrics and applied user experience, with publications in peer-reviewed conferences and journals spanning SN Computer Science, Multimedia Tools and Applications, and ACM Transactions on Computer-Human Interaction.

Through these contributions, Choi has helped contextualize the challenges of bringing Korean language resources into parity with more dominant languages in AI, and has provided key methodologies that have improved Naver’s capacity to capture the unique nuances of Korean communication online.

Naver’s AI-Driven Translation Ecosystem

Historical and Technological Foundations

Naver’s journey to becoming South Korea’s leader in AI-driven translation began as a response to a critical digitalization challenge: the scarcity of Korean digital content and translation resources in the early 2000s5. Through focused investment in NLP research and user-generated knowledge bases, Naver laid the groundwork for a sovereign language processing ecosystem, culminating in the creation of platforms like Papago-South Korea’s most widely used machine translation service, now supporting text, speech, image, and website translation across Korean, English, Chinese, Japanese, and additional languages6.

Recent Improvements and Multilingual Approaches

The release of HyperCLOVA X marked a turning point in Naver’s translation capabilities. Unlike previous generation models, HyperCLOVA X was trained on a balanced, high-quality mix of Korean, English, and code data, followed by instruction-tuning with human-annotated datasets that reflect local language use and cultural expectations7. This model was engineered not just to translate literally but to preserve and infer subtle contextual cues inherent in Korean language communication. Its design addresses critical weaknesses in English-centric systems, which often fail to grasp honorifics, politeness levels, or culturally specific syntax.

Crucially, Naver’s translation systems have embraced multilingual and multimodal innovations. In addition to text-based translation, Naver’s research arms-including Naver Labs Europe-have published acclaimed results in speech translation challenges, outperforming previous state-of-the-art models for low-resource languages like Tamasheq and Quechua at the IWSLT 2023 competition, and demonstrating highly competitive results on downstream language understanding and cross-lingual inference tasks8.

These translation systems harness advances in efficient model scaling, parameter-efficient training, non-autoregressive decoding for speed, and robust user-participatory data enrichment, all of which help bridge the accuracy gap with Western competitors9.

Naver’s AI Content Generation Platforms

Feature Set and User Focus

Naver’s ambitions in AI content generation center around empowering Korea’s vast ecosystem of creators and professionals. Its flagship platforms, including CLOVA Studio and a suite of AI tools for Naver Webtoon and blogging, are driven by generative models that produce and refine text, video, and even avatar-based content for a plurality of end users10.

A landmark feature is the recent integration of AI tools such as AI Painter for comics, MUAi for understanding video context and generating personalized video timelines, and AutoClipAi for converting text (blogs, reviews) into compelling video clips. Livestreaming, a core driver of Korea's internet culture, is supported by advanced avatar-based streaming with speech-to-text and AI-scripting features that process and summarize content in real time10.

These tools are not merely add-ons, but foundational to the new Korean digital creative economy. Naver enables creators to scale content production, expedite editing, and access automatic translation-all of which are crucial to competing globally in sectors such as K-pop, e-sports, and webtoons.

Integration and Monetization

Naver’s fusion of AI generation with its platform ecosystem represents a distinct philosophical and commercial model compared to Google or OpenAI. It prioritizes creator attribution and monetization-a deliberate strategy to reward localized, community-driven content and foster a self-sustaining ecosystem11. AI-generated recommendations, seamless integration with commerce tools, and services like FeedMaker and Prism Live Studio have led to measurable increases in creator engagement and time spent on the platform.

The resulting synergy between AI tools and platform incentives has doubled the number of active creators and led to double-digit increases in user engagement, as measured by time-on-site and active participation in AI-enhanced creative processes.

Strategic Direction of Naver’s AI Initiatives

Embracing “Sovereign AI”

Naver’s strategic direction is defined by its explicit embrace of “sovereign AI”-the vision that a nation’s language models, data infrastructure, and regulatory frameworks should be self-reliant and culturally grounded. This is a response to global shifts, including increased U.S.-China rivalry and the realization among Korean policymakers that local language, culture, and data security cannot be outsourced to international tech giants12.

CEO Sooyeon Choi and leadership figures like Choi Seung-woo and Ha Jung-woo have openly advocated for targeted AI investments, high-performance data centers such as the GAK Sejong complex, and the open-sourcing of smaller, efficient models to stimulate domestic innovation13.

Naver leads this strategy through a multi-pronged approach: massive R&D expenditure (roughly 20-25% of revenue), active participation in government-backed GPU procurement and AI infrastructure projects, and university and industry partnerships that consolidate the domestic AI talent pipeline while promoting open research and commercialization for the Korean market14.

Focus on Vertical, Domain-Specific Solutions

Unlike Google and OpenAI, whose models are optimized for broad, general-purpose utility, Naver centralizes its AI efforts in verticalized, industry-specific applications. This includes custom solutions for the Korean education sector (via partnerships with Daekyo and NSDevil), high-precision translation for legal and medical contexts, and field-specific platforms such as WhaleSpace for academic assessment and digital classroom management15.

Naver’s approach also incorporates strategic penetration of international niche markets. Instead of direct competition in the U.S. or Europe, Naver focuses on extending sovereign AI technologies to other culturally and linguistically distinct markets, such as Japan (via LINE WORKS), Saudi Arabia, and Southeast Asia, often in tandem with partners like Aramco Digital or local education enterprises.

HyperCLOVA X: Development and Capabilities

Architecture and Training Paradigm

HyperCLOVA X is Naver's flagship large language model, representing a step-change in the depth, breadth, and efficiency of AI capabilities in the Korean ecosystem. Built on a compute-memory-balanced Peri-LN Transformer architecture, HyperCLOVA X was pretrained on a strategic corpus of approximately six trillion Korean and English tokens, augmented with high-fidelity synthetic Korean data to ensure domain and cultural specificity16.

The training pipeline employed novel techniques such as curriculum expansion (increasing context window to 128k tokens), supervised fine-tuning, and reinforcement learning from verifiable rewards (RLVR). This careful approach allowed the model to deliver highly competitive performance in bilingual and cross-lingual benchmarks, robust machine translation, and nuanced reasoning and knowledge tasks. Notably, the model’s curriculum ensured that even as training scale increased, Korean-language performance did not degrade-a key concern for non-English languages with limited representation in global AI corpora.

Performance Benchmarks and Multilingual Features

On standardized Korean AI assessments such as KMMLU (Korean Massive Multitask Language Understanding)-which tests 45 domains, including both universal and Korean-specific questions-HyperCLOVA X outperformed both OpenAI’s GPT-3.5 Turbo and Google’s Gemini Pro, and even surpassed GPT-4 in tasks measuring Korea-specific knowledge17, 7. The model was also shown to perform competitively in multilingual inference, handling low-resource Asian and African languages with minimal additional training-a testament to its architectural efficiency and high-quality data pipeline9.

HyperCLOVA X’s multilingual competence extends beyond simple translation, supporting end-to-end cross-lingual inference, multimodal input (text, images, audio), instructional tuning, and even domain-specific verticals (e.g., legal, educational, commerce). Its lightweight variants, such as HyperCLOVA X SEED 14B Think, employ aggressive pruning and knowledge distillation to deliver comparable reasoning performance at a fraction of the training cost and with on-device compatibility for edge infrastructure-showcasing both scalability and flexibility18.

Open-Sourcing and Ecosystem Impact

A distinctive element of Naver’s AI ecosystem under Choi Seung-woo's direction is the commitment to open-sourcing key models and GPU infrastructure. Smaller versions of HyperCLOVA X were released for free via platforms such as Hugging Face, with accompanying reasoning models slated for further public release-a move designed to “seed” the domestic AI research and startup sector and reduce dependence on imported, often opaque, international systems19.

The company’s scope of open source contribution-from language modeling to software engineering for multi-GPU orchestration-has emboldened South Korea’s capacity to independently develop, benchmark, and deploy advanced AI systems across sectors ranging from healthcare to finance to creative industries.

Naver vs. Google and OpenAI: Comparative Analysis

Language and Cultural Proficiency

Naver’s main competitive advantage over Google and OpenAI lies in Korean language processing and cultural context. While Google Translate and OpenAI’s GPT models provide impressive results for major languages, they are fundamentally limited by training data that underrepresents non-English contexts. HyperCLOVA X, with its Korean-optimized tokenizer, deep contextual grounding, and locally sourced training data, produces more accurate, context-aware translations, and generates content that reflects societal norms in Korea, such as honorifics and formality levels9, 7.

KMMLU benchmark results further corroborate this, showing HyperCLOVA X ahead in “Korea-specific knowledge” domains, where translation nuance and local expertise matter most. North American models lack this fine-tuned cultural sensitivity, often making translation errors or missing meaning in domain-specific Korean scenarios17.

Model Scaling, Vertical Integration, and Open Source

Whereas OpenAI and Google have pursued ever-larger general models (e.g., GPT-4, Gemini) with global language coverage, Naver’s strategy is vertical integration, embedding Korean sovereignty throughout the data, model, and service stack. The vertical model means Naver can tightly couple its AI with platform user data, optimizing content personalization, and monetization-whereas Google remains more dependent on open-web indexing and advertising.

Further, by open-sourcing lighter model variants and providing infrastructure support, Naver distinguishes itself as an infrastructure-as-a-service provider for Korea’s broader AI industry, while Google and OpenAI retain proprietary control over their frontier models. This approach strengthens Korea’s national development capacity through shared technical foundations and resource pooling for startups and research institutions-a policy supported by recent government GPU procurement and ecosystem support measures20.

Translation and Generative Quality

In side-by-side tests and community evaluations, HyperCLOVA X matches or exceeds Google Gemini and OpenAI GPT-3.5 in Korean machine translation, particularly on formal and nuanced texts. Although OpenAI’s GPT-4 generally outperforms Naver in broad, multi-language and reasoning tests, HyperCLOVA X maintains an edge in Korea-specific legal, educational, and cultural contexts-key for sectors requiring high translation fidelity and regulatory compliance21.

Naver’s multimodal and real-time translation tools, already integrated into platforms like Papago and Webtoon, further close the user experience gap compared to Western competitors, who lack deep integration in Korean platform ecosystems.

Naver’s AI Partnerships and Collaborative Initiatives

Naver’s leadership in AI has been augmented by deep partnerships across industry, academia, and government. On the infrastructure side, Naver has collaborated with Intel and top universities (KAIST, POSTECH, Seoul National University) to pioneer AI chip initiatives using Intel’s Gaudi platform, thereby reducing dependence on NVIDIA GPUs and enhancing cost-effective model training and inference performance-recent results show inference speeds up to 1.2x those of NVIDIA A100 chips for certain open-source LLM workloads22.

Academically, Naver Labs Europe co-leads France’s national AI research via the MIAI and PRAIRIE institutes, and shares resources and expertise in EU government projects. Naver also maintains a long-standing PhD program and invests in collaborative R&D partnerships aimed at pushing the frontier of machine learning and robotics, demonstrating a strong, transnational commitment to research excellence23.

Domestically, Naver is integral to South Korea’s government-backed sovereign AI projects, serving as the de facto national partner for model and infrastructure deployment. Industry partnerships extend to education (e.g., with Daekyo on WhaleSpace and academic assessment platforms), e-commerce (Wallapop, Poshmark acquisitions for data diversity), and regional industry groups, further solidifying Naver’s centrality in Korea’s AI ecosystem15.

Naver’s AI Infrastructure and Platforms

Data Centers, Compute, and Sovereign Stack

Naver’s AI expansion is underpinned by massive investment in infrastructure, highlighted by the opening of Asia’s largest data center-GAK Sejong-which currently houses the largest GPU clusters dedicated to AI model training in Korea, and is equipped for exabyte-scale storage and next-generation networking24. These hardware foundations are coupled with in-house R&D into neural processing units (NPUs) in partnership with Samsung Electronics, and aggressive government-supported acquisition of high-end NVIDIA H200 and B200 GPUs for industry-scale research and commercial deployment12.

Additionally, Naver’s stack supports GPU-as-a-Service offerings for Korean industry and academia, making high-end compute available for startups and non-corporate researchers. Its software stack-featuring advanced orchestration, accelerated compute libraries, and security protocols-creates a sovereign infrastructure relatively independent from foreign supply chains and platform risks, a point repeatedly emphasized by both Choi Seung-woo and company leadership as well as government advisors20.

Developer and User Ecosystem

Through the open release of the HyperCLOVA X Seed models, comprehensive documentation, and APIs, Naver enables domestic firms to create customized LLM services catering to niche markets (e.g., fintech, healthcare, local government). By making it easier to build and refine Korean-optimized models, Naver lowers the barrier for ecosystem-wide innovation16.

Digitalization of Korean Language and Platform Integration

Naver’s platform strategy is rooted in its early efforts to digitize and structure Korean language data at a scale previously unmatched5. By integrating user-generated content-from Q&A knowledge bases to blog posts and forums-into its search algorithms, Naver cultivated a uniquely Korean web that fosters content localization and vertical integration.

With AI-native tools such as AI Briefing and AI Tab, Naver has begun to transform search and content navigation into an interactive, context-driven experience. Unlike Google’s AI Overviews, which summarize and contain web content within its own interface, Naver’s integration actively credits and surfaces creator content, offering enhanced visibility and monetization for Korean originators and reinforcing the country’s cultural narrative within digital spaces11.

Sovereign AI and Government Initiatives

National Strategy and Investment

South Korea is now investing heavily to become the world’s third AI powerhouse by prioritizing sovereign AI at every level of the value chain-from hardware, infrastructure, and cloud, to foundational model development, deployment, and regulation25. This approach is spearheaded by extraordinary cooperation between public and private sectors, evidenced by Naver’s central role in the government's ₩1.5 trillion GPU procurement, the shortlisting of Naver as head developer for national foundation models, and executive appointments such as Ha Jung-woo (from Naver AI Innovation Center) as the country’s first senior presidential secretary for AI25, 14.

Rather than merely duplicating foreign innovations, South Korea’s “AI for everyone” initiative supports the proliferation of multiple homegrown AI models, customized for diverse industrial and social applications, and provides open access to infrastructure and public datasets for research and commercial use.

Risks and Opportunities

While the sovereign AI push reinforces Korea’s digital autonomy, it also raises questions of competition fairness, technological monoculture, and long-term sustainability-especially as Naver stands poised to “monopolize around ₩2 trillion in government AI contracts,” blurring the line between national policy and corporate aggrandizement12. Policy advocates and industry critics continue to debate the extent to which Naver should represent the national ecosystem, highlighting the importance of fostering a pluralistic AI landscape with meaningful participation from startups, universities, and alternative providers25.

AI Ethics, Regulation, and Policy at Naver

Principles and Practical Mechanisms

Naver has publicly committed itself to responsible, human-centric AI development, introducing its own AI Ethics Principles as early as 2021 in collaboration with Seoul National University’s AI Policy Initiative26. These principles revolve around five concrete standards: “Developing Human-centered AI,” “Respecting Diversity,” “Balancing Explainability with Convenience,” “Ensuring Service Safety,” and “Protecting Privacy and Data Security.” Unlike generic corporate statements, these guidelines are operationalized through CHEC (Consultation on Human-centered AI’s Ethical Considerations)-a process that embeds ethics review into every phase of AI service planning, development, and launch.

This framework, presented at global venues like the UN Office in Geneva, includes interactive ethics consultations, risk assessments (via the AI Safety Framework, ASF), and active participation in open international initiatives such as MLCommons and C2PA for AI content watermarking and provenance standards26.

Addressing Technological and Social Risks

Naver identifies three main risk categories: technological limitations (hallucination, bias propagation, copyright infringement), misuse (AI for misinformation or scams), and loss of human control (overreliance or open source misuse)27. The company’s strategy is a mix of technological safeguards (curated training data, safety-aligned instruction tuning), regulatory compliance (privacy, transparency), and user/community education.

By combining open dialogue with academic and civil society partners, broad user consultation, and technical diligence in design, Naver seeks to lead not only in AI innovation but also in responsibly guiding the rollout and evolution of next-generation digital platforms catering to Korean values and governance expectations.

Table: Naver’s AI Initiatives, Strategic Goals, and Comparison with Global Leaders

The table above outlines Naver’s diverse set of AI-related initiatives, anchored by sovereign large language models and comprehensive creator and user platforms. Naver’s relative strengths lie in local context, vertical platform integration, government partnership, and end-to-end infrastructure, contrasting with the more globally generalized and open-web-centric approaches of Google and OpenAI. The company’s unique positioning as both platform and infrastructure builder for Korea’s digital economy enables it to serve markets and cultural needs often ignored by U.S.-based providers.

Conclusion

Choi Seung-woo’s vision and stewardship within Naver’s AI division have positioned both himself and his company at the forefront of South Korea's digital transformation. Through the development and deployment of HyperCLOVA X and associated generative, translation, and creator tools, Naver has not only advanced the frontiers of Korean-language AI but also established itself as a bellwether for regional, sovereign approaches to artificial intelligence worldwide.

Naver’s model-a fusion of local language innovation, vertical platform integration, open-source collaboration, and robust ethical governance-presents a distinct and increasingly influential alternative to the English-centric, proprietary paradigms of Google and OpenAI. By aligning corporate ambition with national policy, fostering broad ecosystem participation, and prioritizing both technological capacity and cultural resonance, Naver exemplifies the high-stakes experiment in AI autonomy now unfolding in South Korea.

The coming years will test the durability and scalability of this approach against the sheer resource dominance of global big tech. Yet, in translation, content generation, infrastructure, and AI governance, Naver and Choi Seung-woo’s leadership offer a compelling blueprint for countries seeking both digital sovereignty and international competitiveness in the age of generative AI.

References

  1. 1. The state of AI in South Korea and globally, an interview ... - YouTube. https://www.youtube.com/watch?v=2cp2ZWvhFzw
  2. 2. Seungwoo Cho - Research Engineer at Naver. https://www.zoominfo.com/p/Seungwoo-Cho/9328995025
  3. 3. 초거대 AI 모델 및 플랫폼 최적화 센터. https://champ.snu.ac.kr/publications-paper
  4. 4. Seung-Woo Choi - dblp. https://dblp.org/pid/28/7541
  5. 5. Naver CEO Outlines AI Strategy, Claims Third Global AI Model Position. https://www.thepickool.com/naver-ceo-outlines-ai-strategy-claims-third-global-ai-model-position/
  6. 6. NAVER Machine Translation System for WAT 2015 - ACL Anthology. https://aclanthology.org/anthology-files/anthology-files/pdf/W/W15/W15-5008.pdf
  7. 7. Naver's HyperCLOVA X Claims to Outperform OpenAI and Google. https://koreatechtoday.com/navers-hyperclova-x-claims-to-outperform-openai-and-google/
  8. 8. NAVER LABS Europe’s Multilingual Speech Translation Systems for the .... https://arxiv.org/html/2306.07763
  9. 9. Introducing HyperCLOVA X, our state-of-the-art AI models optimized for .... https://clova.ai/en/tech-blog/introducing-hyperclova-x-our-state-of-the-art-ai-models-optimized-for-the-korean-language
  10. 10. Naver unveils generative AI tools to innovate content creation .... https://koreajoongangdaily.joins.com/news/2025-07-17/business/tech/Naver-unveils-generative-AI-tools-to-innovate-content-creation-consumption-for-global-creators/2355323
  11. 11. How Naver’s AI Search Is Reshaping the Korean Digital Landscape. https://www.theegg.com/seo/korea/how-naver-ai-search-is-reshaping-the-korean-digital-landscape/
  12. 19. HyperCLOVAX-SEED-Think-14B - Hugging Face. https://huggingface.co/naver-hyperclovax/HyperCLOVAX-SEED-Think-14B
  13. 12. From Sovereign AI to Naver AI: Is the Line Blurring? - IT조선. https://it.chosun.com/news/articleView.html?idxno=2023092144595
  14. 13. Naver claims its open source HyperClova AI models can rival ChatGPT 4o. https://koreajoongangdaily.joins.com/news/2025-04-23/business/industry/Naver-claims-its-open-source-HyperClova-AI-models-can-rival-ChatGPT-4o/2292121
  15. 14. Tech expert from Naver named inaugural senior secretary for AI. https://www.koreatimes.co.kr/southkorea/politics/20250615/tech-expert-from-naver-named-inaugural-senior-secretary-for-ai
  16. 15. Naver Corp Partners to Revolutionize Education with AI-Driven ... - JAKOTA. https://jakotaindex.com/news-hub/naver-corp-partners-to-revolutionize-education-with-ai-driven-services/
  17. 16. HyperCLOVA X Technical Report - arXiv.org. https://arxiv.org/pdf/2404.01954
  18. 17. Naver says HyperCLOVA X surpasses OpenAI GPT, Google Gemini in Korean .... https://www.chosun.com/english/companies-en/2024/02/27/JYMOZXIYA5HVPLO25IOT5AAKSY/
  19. 18. HyperCLOVA X THINK. https://clova.ai/cdn/media/2025/06/HyperCLOVA_X_THINK_Technical_Report.pdf
  20. 20. Naver, Kakao, NHN to Secure 13,000 GPUs with 1.5 Trillion Won .... https://www.businesskorea.co.kr/news/articleView.html?idxno=248294
  21. 21. HyperCLOVA X surpasses GPT-4 in Korean AI evaluation. https://www.kedglobal.com/artificial-intelligence/newsView/ked202402270003
  22. 22. Intel, Naver, and KAIST achieve AI breakthrough with Gaudi in Korea .... https://biz.chosun.com/