OpenAI expands in Asia with Kakao and Softbank partnerships

OpenAI and South Korean tech firm Kakao, developers of the popular KakaoTalk messaging app, have formed a strategic partnership. This partnership will involve integrating ChatGPT functionality into KakaoTalk, while also providing OpenAI with valuable training and user behavior data. This announcement comes on the heels of Japan’s SoftBank committing $3 billion to implement OpenAI technologies across its various business ventures.

The collaboration with Kakao will encompass three initial projects. Firstly, they aim to create a Korean-language assistant akin to Siri or Google’s Assistant, leveraging OpenAI’s technology. Secondly, Kakao employees will be granted access to ChatGPT Enterprise for their workflows. Lastly, there will be an integration of KakaoTalk with the aforementioned technology.

The agreement with SoftBank follows a similar structure. The two companies will work together on developing an enterprise AI called “Cristal intelligence,” which will securely integrate the systems and data of individual enterprises in a customized manner. SoftBank has also committed $3 billion annually to deploy ChatGPT Enterprise across its subsidiaries and implement Cristal intelligence at scale.

During the announcement press conference, Altman, accompanied by Kakao CEO Shina Chung, praised Korea’s advanced AI adoption. He emphasized the importance of the Korean market to OpenAI due to its rapid growth and strong environment for AI application. Additionally, OpenAI gains access to valuable Korean and Japanese language data for training its language models.

This isn’t the first collaboration between OpenAI and SoftBank. In January, the two companies, as part of a larger consortium, unveiled the Stargate Project. This initiative aims to invest up to $500 billion over the next four years in establishing AI data centers and power generation plants to support OpenAI’s expansion strategy. However, the emergence of China’s DeepSeek AI app, offering comparable performance to OpenAI models at a lower cost and power consumption, poses a challenge to such endeavors.