In a conference room overlooking the hills of San Francisco, executives from two of the world's most important semiconductor companies put their signatures to a document that could reshape how AI infrastructure is built. On 25 July 2026 Samsung Electronics and Broadcom formalised a memorandum of understanding that stretches across memory chips, contract manufacturing and advanced packaging. The partnership is projected to exceed $200 billion in value through 2030.
The agreement is not a binding contract but a statement of intent. Yet its scale reveals something larger about the mechanics of technological advance. Rather than waiting for regulators to set priorities or subsidies to steer investment, two firms with complementary strengths have chosen to align their capabilities. Samsung will supply high-bandwidth memory, including HBM, to power Broadcom's next-generation AI accelerators. It will also manufacture Broadcom products on its 2-nanometre and sub-2-nanometre process technologies, extending even to wireless broadband solutions.
Advanced packaging forms another pillar. The collaboration includes 2.3D and 2.5D integration techniques on Samsung's most sophisticated nodes. These are the quiet, intricate layers of engineering that determine whether an AI system can move data fast enough to justify its enormous power consumption. The accumulated decisions made in such partnerships decide who gets access to faster, more efficient computing years before the results reach British data centres or hospital imaging suites.
AI is driving unprecedented demand for tightly integrated semiconductor technologies spanning memory, logic and advanced packaging. By expanding our collaboration with Broadcom across these critical technologies, we look forward to delivering greater value to customers while advancing the AI infrastructure of the future.
Young Hyun Jun, vice chairman and chief executive officer of the device solutions division at Samsung Electronics, framed the deal in those terms. His counterpart at Broadcom struck a similar note. Charlie Kawwas, president of the semiconductor solutions group, said: "As AI infrastructure continues to scale, close collaboration across the semiconductor ecosystem becomes increasingly important. By combining Samsung's memory and foundry expertise with Broadcom's AI and connectivity leadership, we aim to continue to deliver technologies that power the next generation of AI infrastructure."
The announcement came during an AI summit in San Francisco and sits within a wider pattern of South Korean semiconductor cooperation with American technology companies. That broader set of initiatives carries an estimated total value of around $950 billion. Samsung itself aims to provide integrated solutions that span memory, logic, foundry and packaging for both AI and high-performance computing. In an industry where single points of failure can cascade into global shortages, such diversification matters.
From a British perspective the implications are practical. Faster, more efficient AI accelerators lower the cost of training models, analysing medical scans or optimising energy grids. They reduce dependence on any single supplier or geography. When private capital and technical ambition align without ministerial oversight, the resulting infrastructure tends to reach users more quickly and adapt more nimbly to real needs.
