The Korean Artificial Intelligence Industry Association (KOSA) announced that the 'Korean Full-Stack AI Consortium' has completed a proof of concept (PoC) applying domestic AI technology at Aramco Digital's industrial site in Saudi Arabia.
KOSA stated on September 2 that the consortium has finalized the technology validation based on use cases for industrial AI developed in collaboration with Aramco Digital.
This validation was initiated following a memorandum of understanding (MOU) signed in February between KOSA and the Korean Full-Stack AI Consortium, which aimed to establish a Korean model for full-stack AI capable of processing complex industrial engineering data.
The Korean Full-Stack AI Consortium includes domestic companies across the AI ecosystem, focusing on AI semiconductors, cloud infrastructure, large language models (LLMs), AI application services, and integrated operations management.
A total of seven companies participated in this validation. Megazone Cloud handled the construction and operation of AI and multi-cloud infrastructure, while FuriosaAI and Rebellion focused on AI semiconductors. Upstage and LG AI Research worked on industry-specific LLMs, NC AI managed 3D modeling, and Euracle oversaw LLM, NPU operations, and AI resource management technologies.
The PoC was conducted to verify the applicability of Korean full-stack AI technology in industrial settings by linking the hardware, AI models, services, and operational management capabilities of domestic companies based on the industrial use cases developed with Aramco Digital. The validation was carried out in two phases.
In the first phase of the PoC, the team examined whether AI could interpret 2D engineering drawings and documents and convert them into 3D models, along with reviewing the integrated management capabilities of related technologies.
The second phase expanded the scope to include engineering data-based use cases that could be validated in a short period, building on the results from the first phase. This involved AI interpreting and structuring unstructured drawings and documents, generating 3D models, and validating the entire process from inference based on domestic AI semiconductors to multi-cloud environment configuration and security responses, as well as the integrated operation of AI services and infrastructure.
Joo Jun-hee, chairman of KOSA, remarked, “This validation is significant as it demonstrates that domestic AI companies have collaboratively examined the actual demands of overseas industrial sites by combining their specialized technologies. It is crucial to secure our own AI infrastructure, models, and operational management capabilities from a sovereign AI perspective, as excessive reliance on specific external services can lead to operational risks due to policy changes, sanctions, or service restrictions.”
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.