llms.blog
AI

Kakao Splits Into KakaoAI and KakaoX to Accelerate AI and Messenger Integration

South Korean platform giant Kakao Corp. announced a corporate split that will separate its core operations into two independent publicly traded entities: KakaoAI and KakaoX. The restructuring, approved by Kakao's board of directors, aims to isolate and accelerate the company's artificial intelligence engineering and messaging ecosystem from its broader investment portfolio. Under the spin-off terms, existing shareholders will receive shares based on a net asset book value split ratio of 36% for

axc
axc
1 min
Kakao Splits Into KakaoAI and KakaoX to Accelerate AI and Messenger Integration

South Korean platform giant Kakao Corp. announced a corporate split that will separate its core operations into two independent publicly traded entities: KakaoAI and KakaoX. The restructuring, approved by Kakao's board of directors, aims to isolate and accelerate the company's artificial intelligence engineering and messaging ecosystem from its broader investment portfolio.

Under the spin-off terms, existing shareholders will receive shares based on a net asset book value split ratio of 36% for KakaoAI and 64% for KakaoX. The restructuring is scheduled for completion on January 1, 2027, following a shareholder vote on December 17, with KakaoAI slated to relist on the Korea Exchange on January 27, 2027.

Reorganizing Around Agentic Messaging

KakaoAI will be led by current Kakao CEO Chung Shin-a and will directly control KakaoTalk, South Korea's ubiquitous mobile messaging platform. The new entity is tasked with transforming KakaoTalk into an AI-native interface by integrating conversational foundation models, workflow agents, algorithmic advertising, and conversational commerce. Kakao management set a financial target of 6 trillion won (approximately $4.3 billion) in annual revenue for KakaoAI by 2030.

KakaoX, headed by CA Council Group Investment Strategy lead Kim Do-yeong, will retain Kakao's mobility, fintech, digital payments, gaming, and entertainment subsidiaries.

Kakao Corporate Spinoff Architecture

Strategic Focus Amid Foundation Model Competition

The spin-off reflects increasing pressure on regional platform operators to streamline capital allocation toward AI infrastructure and model deployment. By separating the high-margin messaging and AI service layer from capital-intensive subsidiary operations, Kakao seeks to give KakaoAI dedicated balance sheet agility to fund compute resources, recruit specialized research talent, and deploy proprietary LLMs across consumer and enterprise touchpoints.

Sources

AIBusinessAsiaPlatforms
ShareX (Twitter)LinkedIn

Written by

axc

More to read

  • Agent Memory Engines in Production: Comparing Mem0, Zep, and Letta Architecture, Graph Stores, and Retrieval Latency

    Large language models are inherently stateless functions. Every invocation begins with an empty memory register, relying entirely on the tokens packed into its context window. In production multi-turn applications, naive context management strategies quickly collapse: appending raw conversational logs inflates inference costs and eventually hits hard context limits, while standard semantic vector search (retrieval-augmented generation) lacks temporal awareness, treats contradictions blindly, and

    1 min

  • Cross-Entropy Loss: The Training Objective Behind Every LLM

    Cross-Entropy Loss: The Training Objective Behind Every LLM Cross-entropy loss is the objective function that guides every large language model during training. Despite its ubiquity, the concept is often treated as a black box. This post explains why cross-entropy loss is the natural choice for training probabilistic models of language, and how it connects to fundamental concepts in information theory and statistics. The Gradient of Prediction Error A language model outputs a probability dis

    1 min

  • NVIDIA Researchers Map Cross-Model KV Caches via Linear Ridge Regression to Accelerate Model Cascading Up to 25x

    In multi-agent systems and tiered model cascades, routing requests across different model sizes is a standard architecture for balancing inference cost and generation quality. However, switching models mid-session has historically imposed a severe compute tax: because each neural architecture maintains its own distinct internal representations, the receiving model cannot read the key-value (KV) states generated by the previous model. The target model must recompute the entire conversation histor

    1 min