Samsung Electronics has unveiled plans to invest over 110 trillion won (approximately $73.3 billion) in capital expenditure and research & development for 2026. This represents a 22% increase from the 90.4 trillion won spent in 2025, which was split into 52.7 trillion won on facilities and 37.7 trillion won on R&D. Nearly 90% of that capital expenditure was directed to the Device Solutions semiconductor division.
The massive investment underscores Samsung's aggressive push to dominate two key areas simultaneously: its AI memory business and its foundry operations. A significant portion of the funds will be funneled into expanding production infrastructure at its Pyeongtaek campus, securing additional EUV lithography tools, and building advanced research capacity. The company has already begun mass production of commercial HBM4, the next-generation high-bandwidth memory critical for AI servers.
The announcement gains further significance following confirmation from Nvidia CEO Jensen Huang that Samsung is manufacturing Nvidia's new AI chips. This partnership is seen as a potential turning point for Samsung's foundry business, which has historically been a weaker performer. Samsung's stock rose as much as 5% on the news.
Co-CEO Jun Young-hyun framed this investment as a response to a "rare supercycle" driven by data-center buildouts and accelerating AI investment, expecting strong chip demand through 2026. He noted "the rise of agentic AI is fueling an explosive surge in orders" for both memory and server storage.
The move places Samsung in direct competition with industry leaders. In the high-bandwidth memory (HBM) market, Samsung's HBM4 rollout directly challenges SK Hynix, which currently holds a lead in supplying HBM for Nvidia's chips. On the foundry side, the $73 billion spending plan surpasses rival Taiwan Semiconductor Manufacturing Company (TSMC). Samsung's fab in Taylor, Texas, is nearing a key EUV milestone and could begin risk production in the second half of 2026, providing a more credible U.S. manufacturing base as TSMC guides for roughly 30% revenue growth in 2026 on AI demand.
The AI chip boom is creating a supply crunch for standard memory used in cars and smartphones, as capacity shifts to high-margin AI parts. SK Group chair Chey Tae-won warned this shortage could last four to five years. Samsung's expanded output plan is partly designed to ease this long-term supply pressure.
