KOSA Full-Stack AI Consortium Completes Industrial AI PoC for Aramco Digital
TECHWORLD ·
✦ AI Summary
The Korea Software Industry Association (KOSA) said on the 2nd that the industrial AI technology PoC for Aramco Digital by the 'Korea Full-Stack AI Consortium' has been completed.
The demonstration was a follow-up project to the 'AI Full-Stack Cooperation MOU' signed last February, and seven Korean AI companies built full-stack AI technology linking AI semiconductors, cloud, LLM, and application services and applied it to Aramco Digital's industrial use cases.
The demonstration proceeded in first and second stages and included 2D engineering drawing and document interpretation, 3D model conversion, linkage with domestic AI semiconductor inference, and verification of multicloud, security, and integrated operations.
The Korea Software Industry Association (KOSA) said on the 2nd that the industrial AI technology PoC for Aramco Digital by the 'Korea Full-Stack AI Consortium,' which KOSA operates, has been completed. The demonstration was a follow-up project to the 'AI Full-Stack Cooperation MOU' signed in February. The MOU was signed by KOSA, the Korea Full-Stack AI Consortium, and Aramco Digital.
In the project, Korean AI companies built full-stack AI technology linking AI semiconductors, cloud, LLM, and application services, and applied the technology to Aramco Digital's industrial use cases. The demonstration was carried out in the form of a technology PoC and has been completed.
The focus of the demonstration was on verifying complex industrial engineering data processing. The verification examined whether the AI infrastructure, models, and application services of Korean companies could be connected in a single structure.
Seven Korean AI companies participated in the Korea Full-Stack AI Consortium. MegazoneCloud was in charge of building and operating AI and multicloud infrastructure, FuriosaAI and Rebellions handled AI semiconductors, Upstage and LG AI Research were responsible for industry-specific LLMs, NC AI handled 3D modeling, and EULARK was in charge of LLM, neural processing unit (NPU) operations, and AI resource management.
The target of the demonstration was industrial field use cases jointly identified with Aramco Digital, and the demonstration proceeded in two stages. In the first PoC, the company reviewed the AI interpretability of 2D engineering drawings and documents, examined the possibility of converting the interpretation results into 3D models, and also checked integrated control functions for related AI infrastructure management.
The scope of the second PoC was set to reflect the results of the first and focused on engineering data use cases that could actually be implemented. Accordingly, the company carried out AI interpretation and structuring of unstructured drawings and documents, generation of 3D models based on the interpretation and structuring results, and verification of the linkage between the generated results and the inference process of domestic AI semiconductors.
The demonstration included the infrastructure that powers AI services within its scope, and was conducted in a way that examined not only the AI services themselves but also the infrastructure supporting them. The targets included multicloud environment configuration, security response, and integrated operation of AI services and infrastructure. The focus of the review was on the full-stack structure rather than individual technologies, and the review centered on applicability to industrial sites. KOSA carried out the demonstration.
Through the demonstration, KOSA verified whether the AI semiconductors, cloud, LLM, 3D modeling, and operations management technologies held by Korean companies could be organized into a single system and whether they could respond to overseas industrial site AI demand. Based on this, it is planning a cooperation model that goes beyond supplying individual AI solutions. The cooperation model covers linkage among data processing, inference, service implementation, and infrastructure operation, and it also presents ideas for identifying targets and laying the groundwork for application to overseas businesses.
Cho Joon-hee, KOSA chairman, said the significance of the demonstration lies in the fact that Korean AI companies combined their own specialized technologies to jointly review the real demand of overseas industrial sites. Cho also explained that if AI infrastructure and models become overly dependent on a specific external service, operational risks could arise from policy changes, sanctions, or service restrictions. He added that, from a sovereign AI perspective, securing in-house capabilities in AI infrastructure, models, and operations management is important.
Source: TECHWORLD · Kim Seung-gi
Original: https://www.epnc.co.kr/news/articleView.html?idxno=406419
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.