SK hynix Breaks Ground on Indiana HBM Plant, Moving Production to H2 2029

SK hynix held a groundbreaking ceremony on August 27 for an AI-memory advanced-packaging site in West Lafayette, Indiana. The company says the project exceeds $4 billion, that its cleanroom is scheduled to open in October 2028, and that mass production of next-generation HBM is expected in the second half of 2029 [1]. This is a packaging and test site, not a full relocation of DRAM wafer manufacturing to the United States: SK hynix says wafers made in Korea will be sent to Indiana for advanced packaging and testing before products are supplied to U.S. customers [1]. The ceremony is therefore a geography and execution milestone, not an immediate increase in HBM bits.

The most concrete change is the schedule. In April 2024, the Indiana Economic Development Corporation described an initial investment of more than $3.87 billion on 90 acres and said next-generation HBM was planned for mass production in the second half of 2028 [2]. SK hynix now gives H2 2029. The two public documents are about a year apart. The company has not said whether construction, equipment, customer qualification, product generation, or another factor caused the change. It also has not disclosed annual capacity, first customers, or orders. The date shift cannot therefore be assigned directly to weaker demand or a technical problem.

HBM value depends not only on DRAM dies but also on stacking, interconnection, testing, and customer qualification. Placing those back-end steps in the United States could shorten one part of the path from Korean wafers to U.S. AI customers and put an R&D testbed closer to customers and universities. That is John Wood’s supply-chain inference from the two project descriptions, not a verified delivery benefit. The core wafers still originate in Korea, and public information does not provide packaging yield, tool-installation progress, or customer-qualification status. The site should not be treated as a substitute for Korean front-end capacity.

SK hynix expects roughly 1,000 people during commercial operations and plans an advanced-packaging R&D testbed beside the lines [1]. Those plans show continued project progress, but they are forward-looking company statements rather than shipment evidence. The more defensible conclusion is that, if the H2 2029 date holds, the United States gains an HBM packaging-and-test option for local customers. Until then, usable HBM supply remains chiefly constrained by existing wafers, advanced packaging, yield, and qualification. Moving the date back means this option will miss the originally expected H2 2028 supply window.

The first verifiable checkpoint is whether the cleanroom opens in October 2028, followed by tool installation, next-generation HBM sample qualification, capacity disclosure, and first customer shipments. If the company confirms construction but not qualification or production, the supply-chain effect remains limited. If it also discloses qualified products, stable capacity, and deliveries, the construction milestone can become evidence of regional HBM supply. This is not investment advice.