The United States and Taiwan have struck what officials are calling a historic trade agreement that dramatically expands overseas investment in America’s semiconductor supply chain. The headline figure is massive: a combined $500 billion package tied to US chipmaking, anchored by a major expansion of TSMC’s presence in Arizona. But despite the scale, the deal is not expected to bring the most cutting-edge TSMC chip production to the US anytime soon.
According to details shared through official channels, the $500 billion total includes TSMC’s earlier $165 billion commitment and lifts the overall investment plan to a new level. The breakdown places roughly $250 billion on TSMC, with the rest attributed to Taiwan’s government. The agreement also sets a new tariff rate of 15%, reflecting Taiwan’s push for trade terms closer to what the US has offered partners like Japan and South Korea.
For TSMC, the Arizona buildout is part of a broader strategy to diversify manufacturing and reduce geographic risk, not an exclusive shift away from Taiwan. The company has also expanded or explored expansion in other regions, including Japan and Germany. Still, with the world watching US chip production ambitions, Arizona has become the centerpiece of this latest commitment.
The additional spending in the US is expected to strengthen multiple layers of the semiconductor pipeline, not just wafer fabrication. Plans reportedly include expanding Arizona operations across multiple fabs (often referenced as Fab 1 through Fab 4), adding advanced packaging facilities (AP 1 and AP 2), and establishing dedicated R&D centers aimed at developing domestic semiconductor expertise and talent. That mix matters because advanced packaging and local R&D are increasingly critical to performance, efficiency, and supply chain resilience, especially as AI and high-performance computing demand accelerates.
After the trade agreement was announced, TSMC CFO Wendell Huang discussed the company’s progress and direction in a televised interview, pointing to a growing manufacturing focus in Arizona and suggesting that yield performance between Taiwan and the US is now on comparable footing. That’s an important milestone because yield is one of the biggest hurdles when scaling high-volume chip production at new sites.
Even with those improvements, TSMC is signaling that the most advanced nodes will remain centered in Taiwan for the foreseeable future. The reasoning is straightforward: leading-edge chip development depends heavily on dense collaboration across engineering teams, supply partners, equipment tuning, and process know-how that has been refined over many years in Taiwan. The island’s production lines are more mature, and the talent pool is deeper, which reduces execution risk at the bleeding edge.
There is also a policy constraint that limits how close the US can get to the newest technology. Taiwan’s “N-2” rule requires offshore production to remain two generations behind the latest node available at home. In practical terms, even hundreds of billions in investment doesn’t automatically translate into the newest chips coming off US lines, because the governing framework is designed to keep the most sensitive manufacturing capabilities onshore.
That tension is likely to grow. A large share of TSMC’s customer base is made up of US fabless companies, and many are pushing aggressively toward newer, higher-end manufacturing nodes. As demand rises for next-generation processes such as A16-class technology, the industry will be watching closely to see how long TSMC can maintain strict separation between its most advanced manufacturing in Taiwan and what it deploys abroad, even as American investment and strategic pressure increase.
In the near term, the deal positions the US to gain more semiconductor capacity, more packaging capability, and more local R&D muscle—major wins for supply chain security and industrial policy. But for those expecting immediate access to the world’s most cutting-edge TSMC production in America, this agreement looks more like a powerful expansion of infrastructure than a fast-track to the newest nodes.
