Samsung Electronics and Broadcom have signed a memorandum of understanding (MOU) to deepen their collaboration across memory, foundry manufacturing, and advanced packaging. The two companies expect their partnership to involve spending exceeding $200 billion through 2030. The agreement was announced at an AI summit held in San Francisco, attended by South Korean President Lee Jae Myung and representatives from major AI firms.

The South Korean presidential office described the Broadcom collaboration as part of broader efforts to strengthen the domestic AI semiconductor industry. However, both Samsung and the presidential office declined to disclose annual supply amounts, pricing, shipment schedules, or how the $200 billion would be divided among memory, foundry, and advanced packaging. Without a firm purchase order, the projected figure represents a long-term strategic framework rather than guaranteed revenue. Samsung is one of the few semiconductor companies worldwide that simultaneously develops leading HBM, operates an advanced-node foundry, and offers in-house packaging — a combination that few competitors can match, as most rivals specialize in either memory or logic manufacturing.

Strategic Significance for the AI Supply Chain

Although Broadcom does not manufacture memory chips directly, it has emerged as one of the world's largest designers of custom AI accelerators. Samsung stated that the expanded partnership would support Broadcom's next-generation AI chips through advances in high-bandwidth memory (HBM), foundry production, and advanced packaging — the process of stacking and interconnecting memory and logic dies that determines how effectively data moves within an AI accelerator. Broadcom currently designs Google's Tensor Processing Units (TPUs) and Meta's MTIA accelerators, both of which depend on substantial volumes of HBM.

The agreement carries significance for the broader AI sector because HBM remains one of the most supply-constrained components in AI hardware. The Samsung-Broadcom arrangement could lock in manufacturing capacity that other chip developers may also seek to access.

Demand continues to escalate. According to Seoul Economic Daily, SK Group Chairman Chey Tae-won stated that Broadcom has been persistently requesting what he described as an astonishing volume of memory chips, underscoring the urgency among hyperscale AI clients to secure future supply.

Industry data reinforces these observations. TrendForce's first-quarter 2026 HBM technology report showed SK hynix maintaining its market leadership, while Samsung recorded the fastest recovery, driven by increasing HBM3E shipments and progress toward HBM4 commercialization. SK hynix's lead has been anchored in part by its role as a primary HBM supplier to Nvidia, whose data-center GPU shipments have set the pace for AI accelerator demand. A subsequent TrendForce update reported that Samsung had successfully validated HBM4 and begun shipping it ahead of any other supplier, strengthening its competitive position amid surging AI demand.

Tight Memory Supplies Amplify Timing

Market conditions add urgency to the agreement. Memory costs have risen throughout the year as AI server demand outpaces supply, and analysts do not expect near-term relief.

On July 3, TrendForce projected that DRAM contract prices in the third quarter of 2026 would increase by 13–18% compared to the prior quarter, while NAND flash prices were expected to climb 10–15%, reflecting persistent tightness in the DRAM market.

SemiAnalysis characterized the current landscape as a "silicon shortage phase," in which advanced logic and memory production cannot keep pace with AI infrastructure investment. Under such conditions, companies willing to make long-term commitments gain greater supply certainty, while manufacturers lock their limited capacity into extended contracts.

Samsung's Push to Regain AI Memory Ground

The agreement with Broadcom comes as Samsung works to recover its standing in AI memory after ceding the HBM lead to SK hynix. Samsung stated that the partnership unites its capabilities in memory, foundry production, and advanced packaging to deliver a comprehensive platform for AI customers.

Samsung disclosed several months ago that it had begun mass production of HBM4. On May 29, the company released what it described as the industry's first 12-layer HBM4E samples for major customers. Samsung claimed the new memory achieves transfer speeds of 16 gigabits per second and bandwidth of 3.6 terabytes per second per stack, designed for future AI accelerators.

Samsung's foundry division has also gained traction. SemiAnalysis reported that Samsung Foundry secured Tesla's AI5 and AI6 programs alongside TSMC and entered Nvidia's data-center supply chain.

If the Broadcom memorandum eventually converts into firm production orders, Samsung would add another prominent AI customer to its growing semiconductor portfolio. The South Korean presidential office also noted that Korean and global technology companies announced approximately $950 billion in semiconductor cooperation during the AI Summit, placing the Broadcom agreement among the event's largest strategic commitments.