Samsung Electronics has announced plans to strengthen its presence in the high-bandwidth memory (HBM) market in 2024 in response to shareholder criticism over its lagging performance in the artificial intelligence sector.

Samsung’s semiconductor division head, Jun Young-hyun, stated that the company will begin shipments of enhanced 12-layer HBM3E in the second quarter and plans to produce advanced HBM4 in the second half of the year. On Wednesday, he was officially appointed co-CEO of South Korea’s largest company following the annual shareholder meeting.

Samsung acknowledges that a slow start in the HBM segment led to its lag behind rival SK Hynix. However, the company is determined not to repeat this mistake with next-generation HBM4 and specialized chips. HBM4 is expected to be integrated into Nvidia’s new Rubin GPU architecture, while SK Hynix has already shipped the first 12-layer HBM4 samples to customers.

To gain Nvidia’s approval, Samsung has modified the design of its HBM3E chips. Nvidia CEO Jensen Huang noted at the CES event in Las Vegas that the company is actively working on a new design and rapidly catching up with SK Hynix.

Jun also expects a recovery in the memory market in the coming quarters, driven by increasing demand for AI and mobile devices, which should positively impact Samsung’s financial performance in the second half of the year.

High-speed HBM memory plays a key role in computing and allows chip manufacturers to capitalize on the rapid growth of AI investments. Unlike traditional memory, the complexity of its production makes it more profitable and less susceptible to fluctuations in supply and demand.