Optimizing for AI: Samsung's Struggle to Pass Nvidia's Memory Chip Tests

The quest for technological advancement in artificial intelligence processors faces a hurdle as Samsung Electronics grapples with heat and power consumption issues in its latest high-bandwidth memory (HBM) chips, delaying their approval by Nvidia

by Sededin Dedovic
Optimizing for AI: Samsung's Struggle to Pass Nvidia's Memory Chip Tests
© David Ramos / Getty Images

Samsung Electronics' latest high-bandwidth memory (HBM) chips have yet to pass Nvidia's tests for use in the American company's artificial intelligence processors due to issues with heat and power consumption, three sources familiar with the matter told Reuters.

The problems pertain to Samsung's HBM3 chips, which are the most widely used fourth-generation HBM standard in AI graphics processors (GPUs), as well as the fifth-generation HBM3E chips that the South Korean tech giant and its competitors are bringing to market this year, the sources said.

The reasons why Samsung failed Nvidia's tests have been disclosed for the first time. In a statement to Reuters, Samsung said that HBM is a customized memory product that requires "optimization processes in tandem with customer needs," adding that it is optimizing its products through close collaboration with customers.

The company declined to comment on specific customers. In separate statements after Reuters first published this report, Samsung stated that "claims of failure due to heat and power consumption are not true," and that testing is "proceeding smoothly and according to plan." Nvidia declined to comment.

HBM - a type of dynamic random-access memory (DRAM) standard first produced in 2013 where chips are stacked vertically to save space and reduce power consumption - helps process the vast amounts of data generated by complex AI applications.

As demand for sophisticated GPUs has surged amid the generative AI boom, so has the demand for HBM. Meeting Nvidia's standards - as the company controls about 80% of the global GPU market for AI applications - is considered crucial for the future growth of HBM manufacturers, both in terms of reputation and profit momentum.

Samsung has been trying to pass Nvidia's tests for HBM3 and HBM3E since last year, three sources said. According to two people, the results of the recent failed test for Samsung's 8-layer and 12-layer HBM3E chips came in April.

Successfully meeting Nvidia's stringent requirements remains critical for Samsung's advancement in the competitive HBM market, influencing both its technological reputation and future profitability.