Samsung Electronics denied on Friday that its high-bandwidth memory chips had yet to pass Nvidia's tests for use in the US semiconductor giant's AI systems.
Nvidia's Approval Process Faces Delays
In an exclusive report by Reuters, it has been revealed that Nvidia still needs to approve Samsung's HBM3 processors, which cannot be used in Nvidia's AI processors. The heat and power problems seen in HBM3 chips are also happening with the HBM3E chips.
"The problems affect Samsung's HBM3 chips, which are the fourth-generation HBM standard currently most used in graphics processing units for artificial intelligence, as well as fifth-generation HBM3E chips that the South Korean tech giant and its rivals are bringing to market this year," according to the report.
The results of Samsung's 8-layer and 12-layer HBM3E chips were released in April 2024. While Samsung's HBM3 chips continue to fail, sources claim that SK Hynix began supplying HBM3E chips to Nvidia in March 2024.
Samsung's Commitment to Quality
"We are currently working closely with multiple companies and continuously testing technology and performance," the Korean tech giant said. "We are conducting various tests to thoroughly verify the quality and performance of HBM."
Samsung reiterated its unwavering commitment to enhancing the quality and reliability of its products. This commitment aims to provide consumers with the best possible solutions and instill optimism about the company's future in the semiconductor industry.
Samsung's Strategic Leadership Changes
According to The Korea Herald, Samsung made a significant leadership change on Tuesday as part of its strategic moves in the AI battle.
Co-CEO Kyung Kye-hyun, a seasoned veteran, was replaced as head of semiconductors by vice chairman Jun Young-hyun, a move that underscores the company's strong determination to enhance its future competitiveness.
Last month, Samsung started mass-producing eight-layer HBM3E chips. They aim to reach 12-layer devices in the second quarter.