Samsung-Broadcom Memorandum of Understanding

Detailed close-up of a computer circuit board showcasing electronic components.

Samsung Electronics and Broadcom signed a strategic memorandum of understanding (MOU) on July 24 local time at the San Francisco AI Summit to build core infrastructure for next-generation AI. The deal is valued at approximately $200 billion, or about 290 trillion won, and runs through 2030. Under the agreement, Samsung will provide high-bandwidth memory (HBM), sub-2-nanometer foundry services, and advanced packaging to support Broadcom's next-generation AI accelerators and communications chips. Broadcom chief executive Hock Tan joined Samsung Electronics chairman Lee Jae-yong, NVIDIA chief executive Jensen Huang, OpenAI chief executive Sam Altman, and other technology leaders at the summit hosted during South Korean President Lee Jae-myung's visit to San Francisco.

Samsung's Technology Stack

Samsung plans to bundle HBM4 and the upgraded HBM4E, its sub-2-nanometer foundry process, and advanced packaging into a one-stop supply chain for Broadcom. The South Korean company began mass production and shipments of the industry's first HBM4 featuring a 4-nanometer base die in February, and in May became the first in the industry to provide HBM4E samples to customers. Samsung said the two companies will cooperate at every stage of chip production, from design and fabrication process optimization to advanced packaging and mass production. Samsung Electronics vice chairman and head of device solutions Jun Young-hyun said the partnership reflects a semiconductor industry shift toward solutions that tightly integrate memory, logic chips, and advanced packaging for the AI era.

Broadcom's AI Revenue Trajectory

Broadcom, a leading fabless designer specializing in application-specific integrated circuit (ASIC) processors tailored to individual hyperscalers and cloud service providers, is emerging as a central beneficiary of the trend toward custom AI accelerators outside NVIDIA's GPU dominance. Broadcom's AI chip revenue in the second quarter of fiscal 2026 rose 143% from a year earlier to $10.8 billion, roughly 15.8 trillion won. The company's AI chip revenue in the third quarter is projected to reach $16 billion, about 23.4 trillion won. Hock Tan said Broadcom is collaborating with both Samsung Electronics and SK Hynix on the latest HBM and combined AI chip solutions, and committed to supplying customized semiconductors for Korean AI models and large language model development.

Wider Korean-US AI Alliance

The Samsung-Broadcom MOU is part of a broader package of semiconductor and AI partnerships worth a combined $950 billion announced during the San Francisco AI Summit, according to presidential chief of staff for policy Kim Yong-beom. SK Group separately signed a letter of intent with NVIDIA for a comprehensive partnership worth more than $500 billion spanning AI data centers and AI memory supplies, while expanding its long-term AI memory partnership with NVIDIA and signing new infrastructure agreements with Microsoft, Anthropic, and Amazon Web Services. The deals collectively extend beyond HBM supply into AI semiconductor design, foundry services, data centers, large language models, and physical AI, forming what Korean officials and executives described as an "AI full-stack collaboration network."

Supply Chain and Competitive Implications

The partnership is seen as positioning Samsung as an integrated AI chip company spanning memory and foundry, offering an alternative to the advanced foundry and packaging supply chain long centered on Taiwan's TSMC. Samsung is expected to benefit by diversifying HBM supply channels, reducing dependence on any single customer, and demonstrating sub-2-nanometer yield reliability to attract more foundry orders. Industry sources cautioned that actual results will depend on customer-by-customer product certification, 2-nanometer yields, packaging mass-production capabilities, and final orders from clients. Aju Press noted that the $950 billion headline figure is cumulative across the planned MOUs and long-term supply collaborations over the next five years rather than confirmed immediate investment, with subsequent contracts needed to specify supply volumes, pricing, production schedules, and foundry terms.

Share this article

FacebookX

3 sources

Sources