Google’s AI Strategy in Korea: Gemini’s Growth and Full-Stack Approach

Early Stage of Gemini Adoption and Market Context

Google’s AI product director Simon Tokumine, speaking at the Google for Korea 2025 press conference, responded to concerns that the Gemini AI model’s Korean market share lags behind competitors . He stressed that Gemini’s rollout in Korea is still at an “early stage” with “much room for growth” .  Industry data supports this: a Statcounter report shows ChatGPT holding roughly 80% of South Korea’s AI chatbot market in mid-2025, while Gemini accounted for only about 3% .  By contrast, Google CEO Sundar Pichai announced that the Gemini app had surpassed 400 million monthly active users worldwide by mid-2025 .  Tokumine noted that we are in a “third boom” of generative AI where many new products will emerge; he emphasized that long-term survival and continuous innovation are more important than early market dominance . In summary, Google acknowledges that domestic adoption is initially slow, but it is focused on building capabilities over the long term  .

Full-Stack AI as a Competitive Advantage

Google highlights its “full-stack AI” approach as a key competitive edge. Tokumine explained that many foundational generative AI technologies originated at Google, and the company’s strength is in “vertically integrating AI to provide the entire stack” . This means Google develops AI capabilities across the entire ecosystem—from research and software frameworks to custom hardware (GPUs/TPUs), cloud services, and final products.  DeepMind’s Manish Gupta echoed this, noting that serving billions of users gives Google deep insight into user needs  . In practice, Google can co-design its AI models with the chips they run on and the cloud infrastructure that delivers them, potentially yielding performance and efficiency advantages that competitors lack .

AlphaFold and Gemini’s Multimodal Innovations

Google also showcased AI research breakthroughs that leverage its full-stack strategy.  One example is AlphaFold, DeepMind’s protein-structure prediction model. Gupta reported that about 65,000 researchers worldwide use AlphaFold, and tasks that previously took days of analysis can now be done in “a few seconds” . This illustrates how Google’s AI can drastically accelerate scientific workflows. Gupta also emphasized that Gemini itself is a true multimodal model – it can understand text, audio, images, and code . He noted that the Gemini 2.5 Pro model has become powerful enough to serve as a personal assistant or “agent,” not just a basic chatbot . These advances – from AlphaFold’s scientific impact to Gemini’s expanded capabilities – demonstrate Google’s intent to push AI far beyond current limits .

New Generative AI Tools and Creativity

Google introduced several new AI-powered tools for content creation . For example, Veo3 is a text-to-video generation model that can create video clips from written prompts, and Lydia is an AI that composes original music .  Addressing concerns that AI might replace human creativity, Gupta stressed that AI should augment, not supplant, human artists. He explained that AI “allows artists to attempt various things they couldn’t do before” and should serve as an inspiring tool rather than a replacement . In practical terms, other features were highlighted as well:
• Veo3 (Video from Text): A model that converts natural language prompts into video content .
• Lydia (Music Composition): An AI tool that composes original music based on user input .
• NotebookLM (AI Research Assistant): A service that ingests large text documents and can reformat them into conversational audio or video summaries .

These tools illustrate Google’s push to broaden AI applications across creative and professional domains.

Long-Term Vision and 10× Innovation

Beyond specific products, Google emphasized a 10× improvement philosophy. Gupta noted Google’s internal ethos that it is not enough to seek small (20–30%) gains; instead, the company strives for innovations that deliver an order-of-magnitude improvement . For instance, NotebookLM can take uploaded text and automatically generate a two-person podcast dialogue or a video overview of that content . These features exemplify Google’s vision of transforming workflows (like research and learning) through AI. Tokumine and Gupta’s remarks underline that Google’s strategy in Korea is focused on leveraging its integrated AI stack and aiming for breakthrough capabilities – even if immediate market share gains are modest  .

Sources: Analysis based on Google’s statements at the “Google for Korea 2025” event  , market share data , and reports on Google’s AI products  . Each claim is supported by cited sources.