Samsung Electronics struck a memorandum of understanding with U.S. chip designer Broadcom in San Francisco, creating a strategic partnership estimated to exceed $200 billion through 2030 to supply high-bandwidth memory, custom AI accelerators, and sub-2-nanometre foundry services.
The semiconductor industry is currently navigating a severe global supply crunch driven by surging demand for artificial intelligence data centres. In response, major technology firms are moving away from relying solely on general-purpose graphics processors and are instead investing heavily in application-specific integrated circuits designed for specialized computational tasks.
San Francisco Agreement and the $200 Billion Scope
Samsung Electronics and Broadcom announced their landmark memorandum of understanding during an AI Summit held at The Midway in San Francisco. Representatives from the Korean government, along with senior executives from both companies, gathered for the signing. Attendees included Jinman Han, President and Head of Samsung Electronics’ Foundry Business, and Hock Tan, President and CEO of Broadcom, as Samsung’s global newsroom reported.
The collaboration is projected to exceed $200 billion across memory and foundry technologies over the next five years through 2030, according to CNBC. However, financial analysts and market observers note that because the arrangement is structured as an MOU rather than a finalized binding order or purchase agreement, the $200 billion figure represents a long-term strategic projection rather than secured, immediate revenue.
Technology Integration Across Memory and Foundry
The partnership brings together Broadcom’s expertise in designing custom ASICs—such as Google’s Tensor Processing Units and Meta’s MTIA accelerators—with Samsung’s vertically integrated manufacturing pipeline. Under the terms of the agreement, Samsung will supply next-generation high-bandwidth memory products designed to feed data directly into Broadcom’s AI processing cores. This builds on Samsung’s recent rollouts of high-density memory samples capable of reaching massive bandwidth per stack.
On the manufacturing side, Samsung’s foundry division will produce Broadcom’s next-generation communications chips and wireless broadband communication products using sub-2-nanometre process technologies. The collaboration also extends to advanced packaging methods built on Samsung’s 2nm process, including 2.3D and 2.5D integration, which stack memory closer to logic to squeeze out higher performance and energy efficiency.
“The expanded collaboration reflects Samsung’s focus on supporting customers with end-to-end semiconductor technologies across an increasingly diverse range of AI and high-performance computing applications.”
Samsung Electronics, via CNBC
Challenging TSMC and Reshaping the Silicon Supply Chain
For Samsung, securing Broadcom’s custom AI portfolio serves as a major commercial validation of its 2nm gate-all-around process node. Industry leader TSMC currently commands the lion’s share of the global foundry market, but Wccftech noted that Samsung’s ability to offer an integrated pipeline combining advanced logic, in-house DRAM, and packaging under one roof gives it a distinct flexibility advantage over pure-play foundries that must coordinate across multiple external suppliers.
By pairing its 2nm node with advanced memory production, Samsung can co-optimize physical chip design, power delivery, and thermal performance directly for clients like Broadcom.
Broader Market Pressures and Regional Impacts
The deal arrives against a backdrop of tight global silicon supplies. Industry forecasts indicate that shortages across advanced memory are likely to persist, keeping upward pressure on capital expenditures for AI infrastructure.
The announcement also coincides with a broader diplomatic and economic push by South Korea. The Week reported that the South Korean government has framed the Samsung-Broadcom accord as part of a larger, roughly $950-billion semiconductor cooperation initiative involving global technology partners. While long-term supply agreements between major designers and manufacturers help stabilize supply chains for hyperscale cloud buyers, analysts point out that smaller regional cloud providers and device makers may still face elevated costs and extended lead times as global wafer capacity remains heavily tilted toward premium AI hardware.
Worth a look
