Tektronix TIF 2026
This page was machine-translated and may differ from the original. View original

KT Forms Alliance with Domestic AI Semiconductor Startup 'Rebellion'

Google 우선 소스Published2022.07.11 10:21

▲ Infographic on KT's Strategy to Secure Full-Stack AI Semiconductor

Domestic AI Startup 'Rebellion' Receives 30 Billion Won Investment… Securing National AI Competitiveness
Aiming to become an 'AI full-stack' provider by applying proprietary AI semiconductors in 'GPU farms'


KT is partnering with domestic AI semiconductor startup Rebellion (CEO Park Sung-hyun) to secure national AI semiconductor competitiveness with the goal of becoming an AI full-stack provider.

KT announced on the 6th that it will make a strategic investment worth 30 billion won and engage in business cooperation with 'Rebellion'. Rebellion is an AI fabless startup that has solidified its position in the AI semiconductor field by discovering development talent and strengthening its competitiveness in AISC design.

■ Fostering a 'Korean NVIDIA'... Joining Hands with Domestic Startups

NVIDIA's GPU market share in the field of building computing infrastructure required for AI service development worldwide approaches 80%. This is because AI services and solutions are developed based on 'CUDA' software provided by NVIDIA, and if CUDA is not supported, there were limitations in utilizing AI computation on GPUs.

*CUDA (Compute Unified Device Architecture): Software developed and provided by NVIDIA that is necessary to utilize GPUs (Graphics Processing Units) for AI computation and development.

KT announced that it will strengthen national competitiveness and secure future business opportunities by entering the AI semiconductor field in earnest through cooperation with domestic startups, overcoming dependence on foreign GPUs, and securing mid-to-long-term AI capabilities.

Last year, KT Cloud applied its proprietary AI framework capable of supporting CUDA to 'Hyperscale AI Computing (HAC),' a pay-as-you-go GPU service launched by the company. Through this, it supports the provision of the same development environment for GPUs from other companies, and enables the application of domestically produced AI semiconductors to Hyperscale AI Computing through integrated development work once they are commercialized. Last year, the company also made a strategic investment in MOREH, an AI infrastructure solution company.

KT plans to bring Rebellion into the fold to serve as the control tower for the AI semiconductor business, focusing on next-generation AI semiconductor design and verification, as well as collaboration on large-scale language models. It was announced that AI infrastructure and application services (KT), AI semiconductor driving software (More), and AI semiconductor capabilities (Rebellion) will converge to build a 'GPU farm' with thousands of GPUs, and in 2023, plans to integrate AI semiconductors developed in-house for hyperscale AI computing into the GPU farm.

They stated that their ultimate goal is to compete with global companies aiming for AI full-stack solutions, such as AWS and Google, and to emerge as a domestic 'AI full-stack' provider that offers both hardware and software using purely domestic technology.

Neural Processing Units (NPUs), optimized for AI algorithms, are cited for their advantages of more than three times the energy efficiency and lower implementation costs compared to GPUs. KT added that it plans to pioneer the NPU market in multiple areas, such as data centers and autonomous driving, and utilize them in various fields currently being pursued by KT, including mobility and financial DX.

KT CEO Ku Hyeon-mo emphasized, “As AI semiconductors are a core area that can become Korea’s next-generation growth engine, I look forward to Rebellion, a leader in the domestic AI semiconductor field, becoming a global fabless company like Nvidia and Qualcomm through collaboration with KT,” adding, “Although the investment environment is becoming more challenging, KT’s investment in excellent startups will continue to expand.”
본 기사에 대한 정정·반론·추후보도 청구는 보도 청구 안내를, 그간 게재된 보도문은 정정·반론보도 모아보기를 참고해 주세요.
김예지 기자
김예지 기자