Samsung Electronics has secured a monumental five-year chip manufacturing partnership with Broadcom, valued at more than $200 billion, a move that significantly intensifies the battle for dominance in the artificial intelligence semiconductor supply chain. The agreement, announced on Saturday, will see Samsung manufacture a range of advanced chips for the U.S. chip designer using its cutting-edge 2-nanometer and sub-2nm process technologies, while also supplying next-generation high-bandwidth memory.
The pact, formalized through a memorandum of understanding, runs through 2030 and represents one of the largest single customer wins in Samsung’s history. It underscores the South Korean giant’s strategy to leverage its unique position as an integrated device manufacturer—offering everything from memory and logic chips to advanced packaging under one roof—to lure major clients away from pure-play foundry rivals like Taiwan Semiconductor Manufacturing Company.
“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 said in a statement.
Under the sweeping agreement, Broadcom will tap Samsung’s 2nm gate-all-around (GAA) node for its core products, including high-speed data communication chips. Samsung will also supply its HBM4 memory for Broadcom’s AI accelerators, with a roadmap to eventually ramp up production of its next-generation HBM4E memory to boost competitiveness in the application-specific integrated circuit (ASIC) market.
The partnership goes beyond simple wafer fabrication. Samsung is expected to provide extensive support from the design and optimization stage through to advanced packaging and volume production. The collaboration will extend to Samsung’s 2.3D and 2.5D integrated circuit packaging technologies, aiming to deliver more powerful and energy-efficient AI and networking chips.
This comprehensive approach is a key differentiator in the escalating foundry war. While TSMC’s 2nm process reportedly contributed to 3 percent of its revenue in the third quarter of 2026, the Taiwanese giant’s pure-play foundry model means it must rely on external memory suppliers to offer a similar integrated solution. Samsung’s control over both its 2nm node and 1c DRAM production allows it to co-optimize a chip’s physical design, power delivery, and thermal performance, reducing integration risks and supply chain friction for partners like Broadcom.
The deal lands at a pivotal moment. Technology companies are increasingly developing their own custom AI accelerators, moving away from a sole reliance on general-purpose graphics processors. This shift is fueling demand for specialized chip design and manufacturing partnerships. For Samsung, the long-term production commitments from Broadcom, one of the world’s leading custom AI chip designers, are expected to significantly boost utilization at its advanced manufacturing facilities.
Samsung has been aggressively courting major technology customers to close the gap with TSMC. Last month, co-CEO and chip division head Jun Young-hyun disclosed discussions with Nvidia CEO Jensen Huang regarding next-generation foundry cooperation, including future HBM4E and HBM5 memory products. Last year, Samsung also secured a $16.5 billion contract to manufacture logic chips for electric vehicle maker Tesla.
The broader context is a national push by South Korea to dominate the AI chip landscape. The country aims to double its memory production capacity within five years and secure world-class manufacturing capabilities, with Samsung and SK Hynix accelerating their global pursuit of AI chip orders.