Samsung Electronics is set to adopt chip-making technology championed by its rival SK Hynix in a bid to compete in the flourishing market of high-end chips driving artificial intelligence (AI), Reuters news report said.
The move comes as Samsung, the leading memory chipmaker globally, aims to catch up in the escalating competition to produce high-performance chips crucial for AI applications. Despite the surging demand for high bandwidth memory (HBM) chips, Samsung has notably lagged behind peers like SK Hynix and Micron Technology in securing deals with AI chip leader Nvidia to supply the latest HBM chips.
Industry analysts attribute Samsung’s setback to its adherence to a chip-making technology known as non-conductive film (NCF), which has encountered production challenges. In contrast, SK Hynix has adopted the mass reflow molded underfill (MR-MUF) method to overcome the weaknesses associated with NCF, enabling it to advance in the AI chip market.
Recently, Samsung has issued purchase orders for chip-making equipment tailored for the MUF technique, signaling a strategic shift in its approach. This move signifies Samsung’s recognition of the need to enhance its HBM production yields, with sources describing it as a necessary step, albeit one involving a degree of humility for Samsung as it follows SK Hynix’s lead in technology adoption.
Responding to queries, Samsung affirmed its commitment to its NCF technology, asserting its suitability for HBM products, particularly in its upcoming HBM3E chips. However, after the publication of the article, Samsung refuted rumors about the adoption of MR-MUF in its HBM production.
HBM3 and HBM3E represent the latest iterations of high bandwidth memory chips pivotal in processing vast volumes of data for generative AI. Despite Samsung’s claims of a stable yield rate, analysts estimate its HBM3 chip production yields to stand significantly lower than SK Hynix’s.
Sources reveal that Samsung is already engaging with material manufacturers, including Japan’s Nagase, to secure MUF materials. However, mass production utilizing MUF is not expected until next year at the earliest, necessitating further testing by Samsung.
Notably, Samsung plans to leverage both NCF and MUF techniques for its latest HBM chip, reflecting its pragmatic approach to address production challenges and stay competitive in the burgeoning AI chip market.
The competition underscores the growing significance of HBM chips, with research firm TrendForce projecting a substantial market expansion this year, reaching nearly $9 billion amid escalating AI-related demand.
The contrasting technologies of NCF and MUF highlight the evolution within the chip manufacturing landscape, with companies seeking innovative solutions to enhance chip performance and address production complexities. SK Hynix’s successful transition to MUF has positioned it as a frontrunner in supplying HBM chips to Nvidia, consolidating its dominance in the market.
While Micron recently entered the high bandwidth memory chip race with its HBM3E chip, Samsung’s HBM3 series is yet to secure Nvidia’s qualification for supply deals, underscoring the competitive challenges ahead for the South Korean tech giant.
