Taylor Fab 2 will reportedly build high-bandwidth memory base dies exclusively for the two AI chip giants, with production expected in 2030
Samsung Electronics is lining up a second foundry in Taylor, Texas. According to research findings on the plan, its job will be making a very specific chip component for two of the most important names in artificial intelligence: AMD and Nvidia.
The facility, called Taylor Fab 2, is slated to produce high-bandwidth memory (HBM) base dies exclusively for those two companies.
What Samsung is building in Taylor
Groundbreaking for Fab 2 is scheduled for early November 2026. Production is expected to begin in 2030.
The plant will cover approximately 2.7 million square feet. That makes it about 1.5 times larger than Taylor Fab 1, the first Samsung facility at the site, which is still under construction.
In practical terms, the base die is the logic layer at the bottom of an HBM stack. It manages how the memory above it talks to the AI processor.
Samsung’s plan splits the work across two countries. The company intends to make the base dies in Texas and pair them with DRAM core dies produced in South Korea. Together, those pieces form complete HBM stacks.
Current HBM4 technology uses base dies built on a 4nm process. Fab 2 is aiming for nodes of 2nm or even more advanced.
Fab 1, Tesla, and the bigger picture
Fab 2 does not exist in a vacuum. Its older sibling, Taylor Fab 1, is being built to serve a major supply agreement with Tesla.
That deal, covering next-generation AI chips, is valued at approximately $16.5 billion. Fab 1 is expected to reach full operational capacity by the end of 2026, then ramp toward production in 2027.
Put the two plants side by side and a strategy comes into focus. Fab 1 is anchored by Tesla. Fab 2 is anchored by AMD and Nvidia. Samsung is effectively pre-selling its Texas capacity to some of the biggest buyers in American tech before the concrete is fully poured.
The research findings frame the expansion as support for US Big Tech as those AI and computing demands climb.
What this means for Samsung, TSMC, and the AI supply chain
The most obvious implication is competitive. TSMC currently dominates the foundry market, and landing orders from AMD and Nvidia positions Samsung as a more credible alternative.
There is also a geographic angle. Building base dies in Texas adds domestic US capacity for a critical AI component. According to the market analysis in the research, that increased capacity may reduce dependency on Asian producers and support US technological sovereignty.
The hybrid model complicates that picture a little, though. The DRAM core dies will still come from South Korea. So the finished HBM stacks will be a transpacific product, with a key slice of the value made in America and another key slice made abroad.
The near-term checkpoints sit at Fab 1. Whether it hits full operational capacity by the end of 2026 and ramps on schedule in 2027 will say a lot about Samsung’s ability to execute in Texas.
