Dnotitia
Category: AI Infrastructure
Korean AI infrastructure company building a proprietary Vector Data Processing Unit (VDPU) chip and vector database (Seahorse) for long-term memory AI systems. Dnotitia was founded in 2023. The company is led by Moo-kyung Jeong (정무경). Based in Seoul, South Korea. Team size: 101-500. Total funding raised: ~$86M (KRW 125B). Latest round: Series A. Key investors include Elohim Partners (Series A lead), Kiwoom Investment, Starting Line, Maple Investment Partners, Daesung Private Equity, Shinhan Venture Investment, Ulmus Investment, Kolon Investment, HB Investment, Tony Investment, SJ Investment Partners, FuturePlay.
- Founded
- 2023
- Headquarters
- Seoul, South Korea
- Team size
- 101-500
- Total funding
- ~$86M (KRW 125B)
Value proposition
High-speed, cost-efficient vector database and AI storage infrastructure powered by a proprietary semiconductor (VDPU) purpose-built for vector search and long-term AI memory workloads.
Products and solutions
Seahorse (high-performance vector database powered by proprietary VDPU chip), Mnemos (personal edge LLM device), PoliQ, LegalQ (AI legal search agent), VDPU IP (semiconductor accelerator IP for vector databases).
Unique value
First company to productize a dedicated semiconductor accelerator (VDPU) for vector database/RAG workloads, combining chip-level and software-level innovation for AI memory infrastructure.
Target customer
Enterprises deploying LLM/RAG systems, government agencies, healthcare institutions, and AI infrastructure providers needing high-performance vector search.
Industries served
AI Infrastructure, Healthcare (medical imaging AI partnership with Korea University Anam Hospital), Insurance (Finders partnership), LLMOps (VESSL AI partnership), Government/Legal (LegalQ), Semiconductors.
Technology advantage
Proprietary Vector Data Processing Unit (VDPU) — first accelerator IP for vector databases; deep semiconductor expertise from CEO's Sapeon background; national R&D project leadership in AI storage and chiplet-based AI semiconductors.
How they differentiate
Combines custom silicon (VDPU) with vector database software rather than relying on general-purpose GPUs, targeting cost and latency advantages for large-scale RAG/vector search — unlike pure-software competitors (Pinecone, Qdrant, Milvus) that run on commodity hardware.
Main competitors
Pinecone, Qdrant, DataStax/Milvus (vector database space), traditional AI accelerator makers like Sapeon/FuriosaAI in the Korean AI chip market.
Key partnerships
VESSL AI (LLMOps), Korea University Anam Hospital & Advanced Medical Imaging Research Institute (medical AI), Finders (insurance AI), SKOINFO, Korea Investment & Securities and Shinhan Securities (IPO joint lead underwriters), STX Engine (Changwon defense AI data center MOU, Jul 2026), Inticube (AI-native contact center, Jul 2026), Gyeongnam Province / Changwon City / Dinocore / Shinhwa Steel (100MW defense AIDC initiative), multiple Korean fabless semiconductor firms (national AI storage / chiplet R&D)
Notable customers
Korea University Anam Hospital, VESSL AI, Finders (insurance), Korean government agencies (via national R&D AI storage projects).
Major milestones
Founded June 2023 (registry/The VC; CEO 등기 2023-06-28), Seed KRW 14B Dec 2023; Pre-Series A KRW 21B Dec 2024 (cum. KRW 35B), VDPU IP unveiled / tape-out path 2024–2025; Seahorse GS Grade 1 Jan 2026; Seahorse Cloud SaaS Mar 2026, Selected Korea Investment & Shinhan Securities as joint IPO lead underwriters (Feb 2026), Series A closed Apr 21 2026 (KRW 90B, ~$63.4M), LegalQ beta launched May 2026, STX Engine MOU for Changwon defense AIDC (Jul 2026), Inticube partnership for AI-native contact center (Jul 2026), Seahorse AI Storage won FMS 2026 Best of Show AI Application Award (Aug 5 2026)
Growth metrics
9+ major government LLM development project wins; 34+ strategic industry partnerships; ~137–138 employees (The VC / national pension, mid-2026); Seahorse AI Storage FMS 2026 AI Application Award (Aug 2026).
Market positioning
Positioned as a next-generation AI infrastructure/semiconductor company bridging hardware and software for vector search, ahead of a planned Korean IPO.
Geographic focus
Primarily South Korea, expanding internationally (SC25 US conference presence) with sovereign AI ambitions.
Patents and IP
VDPU IP (Vector Data Processing Unit IP) — first semiconductor accelerator IP for vector databases, announced Sept 2025.
About Moo-kyung Jeong (정무경)
Ex-CTO at Sapeon (AI semiconductor company).
Latest news about Dnotitia
More AI Infrastructure companies
Official website: https://dnotitia.com