- Korea aims to secure world-class AI computing infrastructure technologies—both hardware and software—based on domestically developed AI semiconductors
- MSIT selects 17 consortia to lead the 2025 K-Cloud Technology Development Project
- A total of 59 R&D institutions selected: 37 from industry (62.7%), 15 from academia (25.4%), 3 from research institutes (5.1%), and 4 from public institutions and associations (6.8%)
The Ministry of Science and ICT (Minister Yoo Sang-im, hereinafter “MSIT”) announced that it has selected R&D institutions to carry out the first-year projects under the newly launched K-Cloud Technology Development Project utilizing domestic AI semiconductors. The project is now officially underway.
As global competition in AI computing—particularly in AI semiconductors—continues to intensify, Korea has been making meaningful progress. Supported by government-led R&D initiatives such as the Next-Generation Intelligent Semiconductor Technology Development Project, promising domestic AI semiconductor startups have begun releasing low-power, high-performance chips.
The K-Cloud Technology Development Project aims to develop the core hardware and software technologies required to build and operate commercial AI computing infrastructure based on domestic AI semiconductors. The project passed a preliminary feasibility study in June 2023.*
*The project was approved by the 5th Comprehensive Committee for National R&D Projects Evaluation in 2024. It will be implemented over six years (2025–2030) with a total budget of KRW 403.1 billion, including KRW 342.6 billion in government funding.
The project is organized into three strategic domains:
① development of infrastructure and hardware for AI data centers,
② development of computing software for AI data centers, and
③ development of cloud technologies optimized for domestic AI semiconductors.
It consists of a total of 28 detailed R&D tasks, with 17 tasks open for proposals in the first year: five in infrastructure and hardware, nine in computing software, and three in cloud technologies. Through a competitive review by the selection and project review committees, 17 consortia were chosen from 39 applicants based on their technological capabilities and commercialization potential, comprising a total of 59 R&D institutions.
*Planned full task distribution: 8 in infrastructure and hardware, 14 in computing software, and 6 in cloud technologies (total: 28).
A key component of the project is the overall management consortium, which is tasked with integrating the outcomes of all strategic domains and delivering the final project results. This role was awarded to the HyperAccel–Rebellions consortium, composed of Korea’s leading fabless companies (specialized in semiconductor design). The consortium will lead the development of a composable server by integrating technologies from the infrastructure and hardware domains and incorporate the results of the computing software and cloud domains to validate the project’s final output.
*A composable server is a next-generation data center server designed to maximize infrastructure utilization and cost efficiency by dynamically allocating computing resources to diverse AI workloads.
Additionally, the flagship project in the computing software domain—a core component of AI computing infrastructure competitiveness—was awarded to a consortium led by Moreh, a leading domestic company specializing in AI infrastructure software. The project will focus on developing open-source solutions that are vendor-independent and broadly compatible with domestic AI semiconductors, with the goal of establishing an open and collaborative ecosystem.
To ensure openness, the project will actively incorporate feedback from across Korea’s AI semiconductor industry and utilize products from various companies for performance validation. The resulting technologies will be broadly shared with domestic fabless and software firms, contributing to strengthened software capabilities across Korea’s AI semiconductor ecosystem. In addition, the project will collaborate with global open-source communities such as the Unified Acceleration (UXL) Foundation to expand international outreach and lead global trends in AI semiconductor software.
In the cloud domain, the lead consortium includes research institutions such as the Electronics and Telecommunications Research Institute (ETRI) and the Korea Electronics Technology Institute (KETI), along with the Korea Association of Cloud Industry. The project is designed to ensure that its outcomes are not dependent on any single cloud provider, but can be widely adopted across Korea’s AI computing infrastructure ecosystem.
The project also includes top-performing hardware companies such as Dnotitia and Panmnesia, as well as software companies including Lablup, OKESTRO, and CryptoLab. Leading universities such as Seoul National University and Yonsei University have also been selected as principal research institutions. In addition, major AI computing infrastructure operators—NAVER Cloud, NHN, and SK Telecom—will participate as partner institutions, forming a united effort across Korea’s entire AI computing value chain.
MSIT aims to enhance the performance efficiency of domestic AI computing infrastructure to a global top-three level by 2030. Through this project, Korea expects to secure the technologies necessary to build and operate a hyper-scale data center equipped with 10,000 units of high-performance AI semiconductors developed in Korea.
Furthermore, MSIT will operate a dedicated project management office for the K-Cloud Technology Development Project. This office will serve as a central platform for collaboration and knowledge sharing across Korea’s AI computing ecosystem—including AI semiconductors, hardware and software, and cloud technologies.
It will also play a key role in monitoring project outcomes through working groups and expert committees, and will work in coordination with the National AI Computing Center to ensure that the developed technologies can be applied in commercial AI computing infrastructure.
Park Tae-wan, Director-General of ICT Industry Policy at MSIT, stated, “MSIT has actively supported the growth of Korea’s AI semiconductor industry by funding and facilitating R&D, demonstration projects, commercialization efforts, and talent development. This year alone, we plan to invest KRW 242.3 billion, including KRW 49.4 billion in supplementary funding.”
He added, “Now is the time to move beyond localization and establish Korea’s own AI computing foundation—an essential national capability in the era of artificial intelligence. Through the successful implementation of the K-Cloud Technology Development Project, we aim to secure global competitiveness in AI computing at the highest level.”
For further information, please contact the Public Relations Division (Phone: +82-44-202-4034, E-mail: msitmedia@korea.kr) of the Ministry of Science and ICT.
※ Please refer to the attached PDF.