Samsung Dedicates Half of 4nm Foundry Capacity to HBM4 Base Die Production

Samsung Foundry has reportedly allocated more than 50% of its 4nm (SF4) manufacturing capacity to the production of HBM4 base dies. The strategic shift signifies a transition in how high-bandwidth memory is constructed, moving away from traditional DRAM-class processes toward advanced logic nodes to meet the performance requirements of next-generation AI accelerators.

According to industry reports, the total 4nm capacity at Samsung’s Pyeongtaek facility is estimated at approximately 30,000 wafers per month. Under the new allocation, roughly 15,000 of those wafers are now dedicated to HBM4 production. This prioritization suggests Samsung is leveraging its “turnkey” advantage—the ability to handle DRAM fabrication, logic die manufacturing, and advanced packaging entirely in-house—to compete against the established partnership between SK Hynix and TSMC.

The transition to the 4nm logic node is a technical necessity for the HBM4 generation. Unlike previous iterations of HBM, the HBM4 base die must support more complex features, including integrated memory controllers and physical layer (PHY) interfaces. Utilizing a logic-based process like SF4 allows for higher transistor density and better power efficiency, which are critical as memory stacks are integrated more closely with AI processors.

Abstract visualization of vertically stacked memory architecture. — requires more complex logic-based base dies to manage increased transistor density and power efficiency.

The Move to Mass Production

Samsung officially began mass production and shipment of its commercial HBM4 in February 2026. By dedicating significant foundry resources to this product line, the company is positioning itself to supply the high volumes required for the current AI hardware cycle, including architectures designed for large-scale data centers.

As noted in analysis by TrendForce, the decision to commit such a large portion of 4nm capacity to internal HBM4 needs may impact the availability of these lines for external foundry customers. The 4nm node is frequently utilized by mobile chip designers and other high-performance computing clients. This internal prioritization indicates that Samsung views the AI memory market as its most critical growth vector for the 2026 fiscal year.

The move also simplifies the supply chain for customers who prefer a single-source solution. While competitors must coordinate between separate memory manufacturers and foundries to produce a finished HBM4 stack, Samsung’s integrated model allows it to control the yield and timing of the entire assembly process. This is particularly relevant as the industry hits critical thresholds for mass production viability, where the complexity of the 4nm logic die requires precise manufacturing coordination to maintain stable output.