现代汽车集团首次对外开放内部AI技术大会，披露AI转型成果及2029年Saemangeum万卡GPUaaS远期规划

Hyundai Motor Group is sharing the artificial intelligence transformation experience it has built up with Korean industry, aiming to help spread AI adoption across the broader economy.
Hyundai Motor (005380) Group said it held a technology conference, "Softeer Tech Talk Off Day 2026," for IT leaders from Korean companies and members of the media at its Yangjae headquarters in Seocho-gu, Seoul, on the 21st.
Softeer Tech Talk is an in-house technology event where IT employees at Hyundai Motor Group share the results of projects they carried out over the year. It has been held since 2023. The group judged that it had reached a point where it could share the experience and cases from its AI transformation efforts, and opened the event to outside participants for the first time this year.
The event disclosed not only successes but also missteps, how they were resolved and the challenges that remain, so that other companies pursuing AI transformation could draw practical lessons from it.
"AI transformation is not completed by adopting one particular technology. It is a process of continuously searching for answers," said Jin Eun-sook, president and head of ICT at Hyundai Motor and Kia (000270). "The questions Hyundai Motor Group has faced are much the same as those any company pursuing AI transformation carries, so we will share our experience, listen to others and work out solutions together."
In the first panel session, "Preparing for the AI Journey," the group described how it built the foundations for AI adoption, including its collaboration environment, data integration and an environment for using generative AI.
Hyundai Motor Group introduced a work management and knowledge-sharing platform along with a cloud-based collaboration environment, so that information and work records previously scattered across the organization accumulate in a single collaboration environment. It also consolidated responsibility for managing data and IT systems and set up a framework for linking and using data across departments, lowering the barriers to data sharing.
As a result, the group collects data in real time from about 12 million connected cars worldwide, securing roughly 140 petabytes of vehicle sensor data. It also manages business operation data such as customer, quality and sales information from dozens of affiliates through more than 7,000 data pipelines.
On that foundation, the group also moved quickly to build an environment for generative AI. It unveiled an in-house generative AI platform, H Chat, in 2023, and more than 70% of group employees now use it in their work. Since July this year, the group has expanded the scope of H Chat use to include areas involving national core technologies.
In the following session, "Business Results Achieved Through AI," the group shared cases of applying AI to actual operations such as demand forecasting, maintenance and production. Hyundai Motor Group turned the search, chatbot and agent functions used internally into a common platform, so that new services can be rolled out quickly by selecting the functions needed rather than being developed from scratch. AI models that had been run separately at each site were also consolidated.
In the "Future of AI" session, participants discussed the challenges that have emerged as AI use spreads across the company and how the group plans to respond. Hyundai Motor Group is building foundations including an environment for safely using AI tools created by employees and graphics processing unit infrastructure to support the development of physical AI.
Starting this month, the group is introducing H-Launch, a platform that allows work tools built with AI by non-developer employees to be distributed as in-house services and used with colleagues, rather than staying on individual PCs.
The group is also progressively building the capability to operate GPU as a service, or GPUaaS, which provides services to organizations that require large-scale graphics processing unit resources, in order to support the development of physical AI such as autonomous driving and robotics. After expanding the scale and raising the level of automation in the fourth quarter, it aims to run the service using more than 10,000 GPUs at the Saemangeum AI data center from the second half of 2029.