TLDR: A recent CB Insights report highlights Korean startups like Upstage and Dnotitia as frontrunners in shaping the future of global AI infrastructure. These companies are recognized for their innovative approaches to efficiency, orchestration, and abstraction in AI development, with significant advancements in foundation models, vector databases, and dedicated AI chips.
A new report from CB Insights, released on April 24, 2025, identifies Korean startups as pivotal players in shaping the future of global artificial intelligence (AI) infrastructure. The ninth annual ‘AI 100’ list by CB Insights spotlights the 100 most promising private AI companies worldwide, with a notable emphasis on Korean innovators driving advancements in efficiency, orchestration, and abstraction within AI development and deployment.
Among the key Korean companies recognized are Upstage and Dnotitia. Upstage, a leader in AI infrastructure, was specifically acknowledged under the ‘AI Infrastructure – Foundation Models’ category. Sung Kim, CEO of Upstage, expressed the company’s ambition to establish a ‘Korean AI standard’ globally, extending its influence to international markets including the U.S., Japan, and Thailand. Upstage specializes in proprietary document processing and large language models (LLMs), focusing on developing scalable, high-performance models for real-world applications. The company recently secured $45 million in Series B Bridge funding to accelerate its enterprise-grade generative AI and global expansion efforts.
Dnotitia, a South Korean AI and semiconductor solutions startup, also earned a spot on the prestigious AI 100 list for its groundbreaking work in vector databases for AI infrastructure. Chung Moo-kyoung (MK), CEO of Dnotitia, highlighted the company’s ‘fundamentally new approach to artificial intelligence – one that begins with the data itself.’ Dnotitia has launched Seahorse Cloud, an MCP-based RAGOps SaaS vector database, and is developing the Vector Data Processing Unit (VDPU), a dedicated vector computation chip. The VDPU is projected to deliver up to 10 times better performance than software-only systems and reduce the Total Cost of Ownership (TCO) by over 80%, aiming to make low-cost Artificial General Intelligence (AGI) accessible to all.
Manlio Carrelli, CEO at CB Insights, commented on the selections, stating, ‘The AI 100 winners are future movers and shakers, taking big swings that will shape the future of AI.’ He added that ‘This year’s cohort is pushing AI applications from emerging to emerged, working on everything from AI agents for complex processes to security and robotics. They’ve also zeroed in on new and promising technologies that might not be on the radar of many enterprises — multi-agent systems, novel computing and data curation approaches, and physical AI.’
Also Read:
- AI Market Undergoes Reshaping: New Innovators and Specialized Providers Challenge Established Tech Giants
- India’s Generative AI Landscape: Key Players and Driving Forces
The 2025 AI 100 cohort includes 21 startups focused on AI agent platforms and infrastructure, 12 in AI observability and governance, and another 12 in physical AI technology and building blocks. These companies collectively raised $11.4 billion in equity funding over time, with more than $2.3 billion secured in 2025 alone (as of April 16, 2025). Notably, nearly 75% of the recognized companies are early-stage (Series A or earlier), and 27 are based outside the United States, representing 13 countries, including South Korea. The cohort has also established over 315 business relationships since 2023 with major industry players such as NVIDIA, Google, and Uber.
