Pictured, from left: AI DC Virtualization Development Team members Changju Kim, Sangho Shin, Taekyung Kim, Jahon Koo, Hyojin Choi, Jian Li, Sangmin Lee.
SK Telecom (SKT) took home four awards at the GSMA Global Mobile (GLOMO) Awards on March 6, 2025. Among the numerous technologies and services SKT submitted, the high-performance and high-efficiency cloud platform, ‘Petasus AI Cloud’, stood out as the ‘Best Cloud Solution’. We sat down with the AI DC Virtualization Development Team, the creators behind Petasus AI Cloud to find out more.
Revolutionizing AI DC Infrastructure with Petasus AI Cloud
Petasus AI Cloud is a high-performance and high-efficiency cloud platform designed for building and operating AI Datacenters (AI DC). It applies virtualization technology optimized for AI operations to maximize the utilization of computing resources and reduce power consumption and operating costs. In particular, Petasus AI Cloud features an architecture tailored for AI operations, enabling telcos to run their AI DC more efficiently. This allows for more stable operation of AI services and applications. The name Petasus represents both the winged hat of the god Hermes in Greek mythology and the storage capacity unit of Peta bytes (1 quadrillion bytes).
From left: AI DC Virtualization Development Team members Changju Kim; Hyojin Choi; Jian Li; and Jahon Koo, Team Leader.
From left: AI DC Virtualization Development Team members Sangmin Lee, Sangho Shin, and Taekyung Kim.
The AI DC Virtualization Development Team in a meeting. The team is responsible for researching the “virtualization” and the “optimization” of AI infrastructure.
From left: AI DC Virtualization Development Team members Changju Kim; Hyojin Choi; Jian Li; and Jahon Koo, Team Leader.
From left: AI DC Virtualization Development Team members Sangmin Lee, Sangho Shin, and Taekyung Kim.
The AI DC Virtualization Development Team in a meeting. The team is responsible for researching the “virtualization” and the “optimization” of AI infrastructure.
From left: AI DC Virtualization Development Team members Changju Kim; Hyojin Choi; Jian Li; and Jahon Koo, Team Leader.
Petasus AI Cloud utilizes xPUA general term for various processors for AI computing, such as GPU/CPU/DPU/NPU. virtualization and GPU cluster virtualization technologies to flexibly adjust AI computing resources according to demand and deliver optimal performance. Notably, the proprietary virtual network stack and the DPU offloadingData Processing Unit Offloading : A technology that reduces computational load and optimizes system performance by using DPU – rather than CPU or GPU – for tasks such as network, storage, and security in the data center. technology developed by SKT enables high levels of security without performance degradation, while the distributed firewall function enhances security isolation between virtual machines. Additionally, Petasus supports a variety of AI hardware from NVIDIA, AMD, & Intel GPUs as well as Rebellions NPUs & SK hynix AI accelerators, improving the scalability of AI infrastructure.
Petasus AI Cloud implementation enables allocation of the necessary amount of computing resources through GPU virtualization technology, reducing server procurement costs. This allows for accommodating more customers on a single infrastructure, maximizing the profitability of the AI DC. When using multiple GPUs within a virtual machine (VM), applying the NVLinkA high-speed data transfer technology between GPUs developed by NVIDIA. virtualization feature allows for the use of NVLink’s ultra-fast communication to handle AI computations. As a result, it can reduce computation time, which in turn can lead to cost savings in operations.
Set to Evolve into a More Intuitive and Powerful Solution
Petasus AI Cloud’s GLOMO Award 2025 Trophy for “Best Cloud Solution”
The GLOMO Awards, presented by the Mobile World Congress(MWC) organizer Global System for Mobile Communications Association(GSMA), is the most prestigious accolade recognizing the most innovative technologies across various fields in the mobile technology industry. SKT has previously won multiple GLOMO Awards for its innovative and technologically advanced solutions. Notably, Petasus AI Cloud’s win in the “Best Cloud Solution” category marks SKT’s second consecutive victory in this category, following last year’s recognition for ‘Cloud Radar.’
Q. What was the background for developing the Petasus AI Cloud?
Jian Li, AI DC Virtualization Development Team member.
A. Jian Li: Petasus AI Cloud began its journey as an edge cloud solution. In 2020, it pioneered the commercialization of edge cloud globally, addressing the limitations of existing solutions that failed to meet the requirements for high-performance, low-latency, and ultra-lightweight virtualization systems. To overcome these challenges, SKT developed its proprietary EdgeStack engine, which formed the basis of its Mobile Edge Computing (MEC) platform.
In 2021, SKT collaborated with Dell to jointly launch the ‘Petasus’ brand, marking a significant milestone in their partnership. By 2023, SKT had expanded its capabilities by integrating a broadcasting equipment virtualization feature into a Platform as a Service (PaaS) model. This enhancement enabled support for media services, leading to the development of the Petasus Media solution. Leveraging this technology, we successfully introduced Petasus Media to the market, including a sale to Sinclair, the largest U.S.-based broadcast group.”
Q. How was the experience on winning the Best Cloud Solution award at this year’s GLOMO Awards?
Sangho Shin, AI DC Virtualization Development Team member.
A. Sangho Shin: I was excited that Petasus was shortlisted, but I also told myself it wouldn’t be easy winning two years in a row since we already won an award last year. I had confidence in our technology, but not in receiving the award. However, I’m extremely pleased to receive this honor once again. This is great recognition for the technology we developed over the past five years.
Q. How did people respond to the Petasus AI Cloud? And what is in store for Petasus AI Cloud in the future?
A. Jian Li: Many global companies at the MWC25 were interested in the Petasus AI Cloud. A Japanese IT company proposed the joint development of GPUaaS ecosystem, while a U.S.-based global storage company discussed conducting a PoC for integrating its storage solutions. In addition, we discussed potential collaboration opportunities with various global companies, and businesses from Europe and Africa showed significant interest in adopting GPUaaS services.
A. Sangho Shin: Moving forward, our goal is to continuously enhance the Petasus AI Cloud to address the cost and power consumption challenges of AI DCs, ultimately improving AI infrastructure efficiency and reducing operational expenses. Additionally, we aim to incorporate the various customer needs identified at MWC25 to refine our solution to be more intuitive and powerful. We also plan to roll out these advancements to our AI DC business to further boost operational efficiency.
GLOMO 2025 award winners pose for a commemorative photo. Back row, left to right: Dongwook Kim, SKT 6G Development Team Leader; Seung-hyun Yang, SK AI R&D Center Director; John Hoffman, GSMA CEO; Jongmin Lee, SKT Future R&D Lab Director; Seongchan Lee, YATAV CEO; and Jong Hwan Um, SKT ESG Promotion Office Director.
Meanwhile, SKT won an impressive four awards at this year’s GLOMO. In addition to Petasus AI Cloud, the company won awards for AI “APS” (Best Mobile Operator Service for Connected Consumers), AI Waste Sorting Guide “Happy Habit” (Outstanding Mobile Contribution to the UN SDGs), “Meta Forest” (Best Mobile Innovation for Enhancing the Lives of Children and Young People). Moving forward, SKT remains committed to driving innovation in AI data centers and cloud infrastructure, aiming to solidify its leadership in the global market.
