‘AI-Semiconductor Initiative’, intensive investment in 9 technological innovations Realization of ‘Artificial Intelligence G3 Leap, Creation of a New K-Semiconductor Legend’ < Policy < FOCUS < Text of Article

Relevant ministries including the Ministry of Trade, Industry and Energy (Minister Ahn Deok-geun, hereinafter referred to as the Ministry of Trade, Industry and Energy) and the Ministry of Science and ICT (Minister Lee Jong-ho, hereinafter referred to as the Ministry of Science and ICT) announced the ‘AI-Semiconductor Initiative’ at the plenary meeting of the National Science and Technology Advisory Council held at the Presidential Office in Yongsan on the 25th. It was announced that the agenda had been deliberated and resolved.

This agenda was prepared to specify the direction of the ‘AI-Semiconductor Initiative’ announced at the ‘Semiconductor Current Issues Inspection Meeting’ presided over by the President on April 9.

With the development and spread of AI further accelerating, starting with generative AI such as ChatGPT and super-large language models, AI is expected to evolve to the same level as humans within a few years, and is accompanied by enormous market growth. Accordingly, not only Big Tech but also countries around the world are competing fiercely to secure leadership in the AI ​​era.

The AI ​​value chain connects various fields such as AI models, AI semiconductors, cloud, devices, and SW. The key to AI development is for all fields included in the value chain to grow organically.

The AI-Semiconductor Initiative consists of nine major technological innovation tasks derived by analyzing the strengths and element technologies of each sector of Korea’s AI value chain and key implementation tasks to support them. The government plans to join forces with the private sector to actively promote the AI-semiconductor initiative to realize ‘AI G3 leap forward and creation of a new K-semiconductor legend.’

The nine major technological innovation tasks are: First, secure the source technology to lead AI technology hegemony.

We develop next-generation AI core technologies, such as next-generation general-purpose AI (AGI), which can overcome the limitations of existing generative AI and perform human-like capabilities in a variety of fields. In addition, we plan to secure lightweight, low-power AI technology that uses less energy while maintaining existing performance, and ultimately develop AI to the point where it can be freely used on all devices. In addition, we plan to secure AI safety technologies such as explainable AI, AI/cybersecurity technology, and deepfake detection technology so that we can trust and use the rapidly growing AI.

Second, secure a super gap/new gap in AI semiconductors. We plan to secure a super gap in the memory field, where we have strengths, through Processing in Memory (PIM), which applies AI calculation functions to memory. We plan to apply PIM technology to DRAM such as HBM and LPDDR and NVM (non-volatile memory) to increase calculation speed and dramatically reduce power usage.

In addition, we are challenging a new gap by developing low-power K-AP, a Korean-style AI processor. We plan to become the first in the world to commercialize a neuromorphic AI semiconductor that mimics the structure of the human brain, and will continue to enhance the NPU, which has recently entered the commercialization stage. Lastly, we will develop new devices and advanced packaging technologies that can fundamentally change the paradigm of semiconductors.

We plan to establish a Lab to Fab scale-up platform so that new device research results can be applied to actual sites rather than staying in the lab, while stably supporting the development of innovative new devices in the mid- to long-term through large-scale R&D investments. In addition, we plan to actively secure cutting-edge packaging source technology, which has recently become increasingly important.

Third, create an AI-semiconductor HW/SW technology ecosystem. AI supercomputing (K-Cloud 2.0) is promoted to advance the cloud with domestically produced AI semiconductors. The cloud is a platform where servers equipped with AI semiconductors operate as one huge system.

We plan to support R&D and verification so that domestic AI semiconductors can operate efficiently in these large-scale cloud data centers. In addition, we are developing core on-device AI technologies that are expected to accelerate the everydayization of AI. We will develop AI semiconductors and devices to drive on-device AI in limited performance and energy environments, and promote K-on-device AI flagship projects in key industries such as automobiles, machinery, robots, home appliances, and defense to help companies capture the initial market. We plan to provide support so that we can actively target the target.

In addition, we will develop a next-generation open AI architecture and software that efficiently controls domestically produced AI semiconductors and supports their operation in data centers and devices. Through this, we plan to complete an AI-semiconductor ecosystem in which HW and SW are organically linked.

In addition, in order to intensively support the 9 major AI-semiconductor technology innovation tasks as key promotion tasks and promotion systems, we provide comprehensive investment and financial support, foster talent, establish industry and research innovation infrastructure, global cooperation and advancement, and lead the AI ​​code of ethics. to support the overall AI-semiconductor value chain. In particular, we plan to solidify Korea’s AI global leadership by successfully holding the ‘AI Seoul Summit’ in May this year.

In particular, in order to realize the AI-semiconductor initiative, the ‘National Artificial Intelligence Committee’, a pan-national promotion system, will be launched under the direct control of the President to marshal all the nation’s capabilities. In addition, we plan to continue to promote industry-academia-research-government cooperation across the AI-semiconductor value chain to ensure results across the entire industrial ecosystem.