Samsung to adopt SK Hynix's chip making technology for AI chips

Samsung in focus: Samsung Electronics is reportedly planning to embrace a chip making technology favoured by its competitor SK Hynix as it seeks to compete in the high-stakes race to produce advanced chips for artificial intelligence (AI) applications.

The demand for high bandwidth memory (HBM) chips has surged with the increasing adoption of generative AI technologies. Despite being a major player in the semiconductor industry, Samsung has notably lagged behind in securing deals with AI chip leader Nvidia for the supply of the latest HBM chips, unlike its peers SK Hynix and Micron Technology.

One of the reasons for Samsung's delay is its reliance on a chip making technology called non-conductive film (NCF), which has encountered production issues. In contrast, SK Hynix has switched to the mass reflow moulded underfill (MR-MUF) method to address the weaknesses of NCF.

Sources familiar with the matter revealed that Samsung has recently issued purchase orders for chipmaking equipment designed to handle the MR-MUF technique. 

The move marks a departure from Samsung's previous stance and indicates its efforts to enhance production yields for HBM chips.

However, Samsung has denied rumours of adopting MR-MUF for its HBM production, asserting that its NCF technology remains optimal for HBM products and will be used in its new HBM3E chips.

Despite Samsung's efforts, its HBM3 chip production yields currently stand at a lower rate compared to SK Hynix. Samsung is reportedly in talks with material manufacturers to source MR-MUF materials, but mass production using this technology is not expected until next year at the earliest.

Sources also revealed that Samsung plans to utilise both NCF and MR-MUF techniques for its latest HBM chip, highlighting the company's commitment to addressing production challenges and competing effectively in the AI chip market.

Samsung's decision to adopt MR-MUF underscores the increasing pressure it faces in the AI chip race, with the HBM chip market projected to grow this year amid rising AI-related demand.

(With Reuters Inputs)