Samsung Teams Up with TSMC for Nvidia's Custom HBM Chips to Boost AI Performance
Yahoo Finance ·
Samsung Electronics is partnering with Taiwan Semiconductor Manufacturing Company (TSMC) to utilize its 3nm process for producing base dies in Nvidia's next-generation custom high-bandwidth memory chips. According to reports, this marks the first time Samsung has cooperated with TSMC during the HBM manufacturing phase. The collaboration aims to address performance and supply bottlenecks in AI chips as GPU processing capabilities outpace traditional memory speeds. Nvidia is co-designing the base die for its proprietary NVHBM memory with both firms, integrating the memory controller onto the base die to free up GPU space. This innovation delivers up to 30% greater memory bandwidth and a 15% reduction in power consumption compared to standard HBM4E. Industry sources noted TSMC's superior yield of over 90% drove the partnership, while UBS projects global HBM demand to surge 91% next year to 61.5 billion gigabits. Although Samsung is collaborating with TSMC, it continues preparations to manufacture base dies internally for future demands.
AI 시장 분석
Samsung Electronics has initiated a collaboration utilizing TSMC's 3nm process to produce Nvidia's next-generation custom HBM chips. This collaboration is expected to improve memory bandwidth by 30% and reduce power consumption by 15%, contributing to easing supply bottlenecks in AI semiconductors. Investors should closely monitor future partnership progress and changes in HBM market share.
상승 영향
- Semiconductors — HBM performance enhancement and 30% bandwidth improvement through the collaboration between Samsung and TSMC will significantly increase AI chip production efficiency and related companies' revenues.
- AI — As memory bandwidth limits and supply bottlenecks ease, the maximization of Nvidia's next-generation GPU performance and the expansion of AI infrastructure will accelerate.
DYAX 전담 분석
This collaboration is structured for Nvidia to utilize TSMC's high 3nm yield (over 90%) to overcome HBM performance limitations and maximize data transmission speeds. By integrating the memory controller into the base die to secure GPU space and with next-year HBM demand projected to surge by 91% compared to this year, supply chain efficiency will be significantly enhanced.
The bullish scenario is that Samsung and TSMC's collaboration succeeds, resolving AI chip supply shortages and driving earnings growth for both companies, while the bearish scenario is that foundry yield gaps or delays in the process integration phase occur, increasing costs. Key indicators to watch are whether TSMC maintains its 3nm yield and the actual resolution speed of next year's 10% HBM supply deficit projected by UBS.
AI가 생성한 분석으로 투자 자문이 아닙니다.
DYAX Investor Sentiment
Bullish (Long) 53% · Bearish (Short) 47%
554 participants
Related News
- Stocks Rise Pre-Bell as Traders Prepare for Busy Week of Corporate Earnings
- Objections to ICE and OKX’s Tokenized Stock Venue Climb to Nine as Broadcom, Verizon Opt Out
- US Equities Focus on Meta and Tesla Earnings Alongside Treasury Yields and Crude Oil Trends
- Why Is Datadog Stock Moving Higher Today?
- Did Tesla's Fall Catch You Out? What Investors Missed
- JPMorgan Chief Dimon Declares US Economy Thriving Amid Global Market Updates
