[Featured Stock] QRT, Inevitable Benefits from HBM and CXL... The Path to a 'Market Cap of 1 Trillion'

QRT is showing strong performance. The expectation that it will benefit from the increasing demand for artificial intelligence (AI) semiconductors appears to be influencing its stock price. As the era of generative AI fully begins, demand for AI semiconductors that accelerate large models like LLMs with high performance and low power consumption has recently surged, and the reliability and quality assurance of these AI semiconductors are being recognized as important competitive advantages. On the 16th (local time), Nvidia's stock price hit an all-time high on the New York Stock Exchange. The prevailing view is that 2024 will be the inaugural year for the AI chip market to open up.

At 9:53 a.m. on the 18th, QRT was trading at 21,400 KRW, up 18.82% from the previous day.

Minjeong Kwak, a researcher at Hyundai Motor Securities, introduced, "Semiconductors go through hundreds of processes before release," adding, "Reliability evaluation verifies how well essential processes can withstand extreme environments." She continued, "Most semiconductor manufacturers establish their own reliability standards to manage quality," and explained, "Due to outsourcing of reliability evaluations caused by product diversification, QRT has significant potential for earnings growth."

She said, "The increase in research, development, and mass production proportions of new semiconductor-related products requiring high testing difficulty, such as HBM, on-device, NPU, and CXL, inevitably leads to an increase in QRT's sales," adding, "Memory such as HBM, DDR5, and CXL, which are faster than existing DRAM, generate more heat, making reliability evaluation more challenging."

Researcher Kwak stated, "As the defect rate increases due to semiconductor process miniaturization, the importance of reliability evaluation through defect analysis inevitably grows," and added, "We expect that comprehensive analysis to identify product defect causes and propose improvement solutions, as well as independently developed lifespan evaluation equipment for 5G RF semiconductors, will be used by clients for mass production."

She estimated sales of 74 billion KRW and operating profit of 12.8 billion KRW for this year. Compared to last year, sales and operating profit increased by 37.0% and 591.9%, respectively. Researcher Kwak analyzed, "The reason for improved performance is the expanding demand for reliability evaluation in new semiconductors such as HBM, CXL, and NPU," and added, "ASP is also formed at a high level, mainly for high-end products."

Since its listing in November 2022, QRT has been aiming for a market capitalization of 1 trillion KRW. Opportunities are arising as the AI semiconductor era opens. QRT signed an MOU with Rebellion, an AI semiconductor specialized fabless (semiconductor design specialist), in November last year. QRT will be responsible for precise reliability evaluation and comprehensive analysis of AI semiconductors developed by Rebellion. A laboratory (LAB) for verifying AI chips was also established on the second floor of the Gwanggyo Vision Campus together with Rebellion.

Rebellion, which has attracted attention domestically and internationally, recently began performance evaluation for AI semiconductor adoption with IBM. It is rare for a global information technology (IT) company to conduct performance evaluation of domestic AI semiconductors. If passed, the semiconductors will be installed in several data centers of IBM, which provides cloud services worldwide.

The next-generation products developed by Rebellion are planned to be produced in cooperation with Samsung Electronics. Samsung Electronics will manufacture the products using its foundry 4-nanometer (nm) process, and will simultaneously provide the 'HBM3E (5th generation)' memory that will be mounted together. Rebellion and Samsung Electronics plan to complete new product development by the second half of 2024.

Rebellion is developing 'Ribel,' an AI semiconductor for large language models (LLMs) used in supercomputers. It powers large AIs such as OpenAI's GPT, Google's Bard, and Meta's LLaMA. Rebellion aims to surpass Nvidia's super AI semiconductor H100.