Ok-Moonshot area PD Lee Choon-woo, head of the Space Exploration Research Center on the Korea Aerospace Research Institute. Courtesy of Lee Choon-woo
[Editor’s note] Artificial intelligence (AI) is remodeling the paradigm of scientific analysis. As AI intervenes throughout your entire means of formulating hypotheses, designing experiments, and analyzing information, international locations are starting to undertake AI as a core instrument for nationwide science and know-how competitiveness. Starting this 12 months, Korea is totally launching a cross-ministerial challenge known as ‘Ok-Moonshot’ to resolve 12 grand nationwide challenges with AI. We spoke with the 5 program administrators (PDs) overseeing the important thing areas carefully tied to superior science and know-how — area, brain-computer interface (BCI), quantum, nuclear fusion, and small modular reactors — in regards to the present standing and technique of their missions.
“Through K-Moonshot, if we develop and demonstrate core components such as semiconductors, solar cells, and thermal management technologies, I believe we can play the role of a supply chain that provides them to the global market.”
Lee Choon-woo, Ok-Moonshot area PD (head of the Space Exploration Research Center on the Korea Aerospace Research Institute), expects that Ok-Moonshot won’t solely obtain the bold purpose of constructing a Korean-style area information heart, but additionally allow Korea to function a provide chain that gives core elements to area information facilities being constructed by world large tech firms corresponding to SpaceX.
The Ok-Moonshot area mission goals to conduct in-orbit demonstration of core applied sciences for area information facilities by 2030, and to launch and validate a kW-class area information heart able to small-scale energy era and companies by 2035. As PD, Lee oversees your entire R&D course of, from mission planning and analysis to administration.
The Space Exploration Research Center at KARI started analysis final 12 months on foundational applied sciences for giant area buildings, together with modules for business area stations that can substitute the International Space Station (ISS), which is scheduled for retirement round 2030. Space information facilities are additionally a kind of huge area construction, and their energy era and cooling points are primarily the identical as these of an area station. As the analysis scope naturally expanded, this turned a possibility for Lee to use for the Ok-Moonshot PD position.
Lee defined, “I decided to apply because the K-Moonshot space mission happened to be defined around securing and demonstrating foundational technologies for space data centers.”
Space information facilities are rising as an alternative choice to overcome the power limitations of terrestrial information facilities. According to world consulting agency McKinsey, because the AI trade expands quickly, energy demand from information facilities is anticipated to greater than triple by 2030. Power provide, transmission, and cooling have turn into bottlenecks in constructing information facilities on the bottom.
In area, it’s attainable to generate 5 to eight occasions extra solar energy than on Earth, and warmth may be naturally dissipated into area by radiative cooling. Because this presents a approach to overcome the restrictions of terrestrial information facilities, firms like SpaceX and organizations in China are racing to announce plans for area information facilities.
SpaceX has introduced a purpose of deploying AI computing in area that consumes 1 gigawatt (GW) of energy per 12 months by the tip of 2027, and then increasing this capability tenfold yearly to achieve the terawatt (TW) degree. China, for its half, goals to construct an area information heart consisting of two,800 satellites by 2035.
Lee notes that, regardless of astronomical figures corresponding to SpaceX’s terawatt-scale AI computing goal, world competitors remains to be at an early demonstration stage. For instance, the satellite tv for pc launched final November by U.S. startup Starcloud is an ultra-small 60 kg satellite tv for pc that skilled a big language mannequin (LLM) utilizing a single GPU. China, in the meantime, has up to now launched a cluster of 12 satellites in May final 12 months.
Some observers query whether or not Korea’s purpose of demonstrating core applied sciences by 2030 is lagging behind, however Lee stated, “Given that the concept is still at the proof-of-concept stage globally, we are not that late,” whereas additionally noting that they’re contemplating transferring up the timeline according to worldwide traits.
Lee plans to hyperlink the Ok-Moonshot missions in semiconductors and photo voltaic cells by integrating Korea’s strengths — AI semiconductors, high-efficiency photo voltaic cells, and radiative cooling-based thermal administration applied sciences — into the area information heart.
Memory semiconductor applied sciences at Samsung Electronics and SK Hynix are already world-class. The variety of home startups making an attempt to develop AI semiconductors for area information facilities can also be rising. Research on high-efficiency photo voltaic cells utilizing perovskite supplies is gaining prominence in Korea. Perovskite has a better gentle absorption effectivity than silicon and is attracting consideration as a next-generation photo voltaic cell materials. Thermal administration applied sciences that dissipate warmth radiatively into area are additionally thought of a aggressive power.
Lee stated, “If we validate these technologies in space through K-Moonshot, we could also take on the role of supplying core components to the space data centers of global big tech companies.”
The position Lee values most is coordination. He famous, “We have to go through processes where we sometimes persuade, sometimes coordinate, and sometimes make decisions among the diverse opinions of industry, academia, and research institutes,” including, “Stakeholders all have different views on whether to build the space data center as a distributed system based on satellite constellations, or as a centralized system with large structures placed in orbit, and what the business model should be.”
He emphasised, “The ultimate goal is for the state, through K-Moonshot, to secure AI-based technologies and platforms, complete them as commercial products and services, and thereby lay the groundwork for active participation by the private sector.”
Copyright ⓒ DongA Science. All rights reserved.