SK Hynix and Samsung Electronics have secured massive long-term agreements with Nvidia and Broadcom, totaling nearly $950 billion in memory chip supplies. This move highlights the relentless demand for AI hardware powering global data centers.
SK Hynix committed to providing $750 billion worth of high-bandwidth memory (HBM) chips, mainly for Nvidia's AI accelerators. This deal not only extends but significantly scales their existing partnership, reinforcing SK Hynix's role as a key AI infrastructure supplier.
Samsung's $200 billion agreement with Broadcom reflects a strategic effort to strengthen its position in the HBM market. Broadcom, which designs custom chips for major cloud providers, is betting on sustained demand from hyperscale data centers by securing a dedicated memory supply.
These announcements came during South Korean President Lee Jae Myung's visit to Silicon Valley, where he engaged with leaders from Nvidia, OpenAI, and other tech firms. The timing shows South Korea's key role in the global semiconductor supply chain, largely driven by US demand that accounts for up to 90% of Korea’s semiconductor expansion.
While SK Hynix has aggressively expanded its HBM production, Samsung is working to close the technical gap after earlier yield challenges. Their deal with Broadcom suggests Samsung is gaining traction with new customers even as it ramps up production capabilities.
The scale of these deals sends a clear message to investors and the market: Nvidia and Broadcom anticipate strong, multi-year growth in AI infrastructure needs. Broadcom’s focus on custom AI chips for cloud giants like Google and Meta further emphasizes this trend.
This content is for informational purposes and does not constitute financial advice.
