AI news ·
AMD pledges $10 billion to Taiwan semiconductor ecosystem to expand AI infrastructure
AMD is investing $10 billion in Taiwan to build AI infrastructure through advanced chip packaging and manufacturing partnerships. Rack-scale systems using its Helios platform are set for deployment in the second half of 2026.
AMD Commits $10B to Taiwan for AI Infrastructure Development
Advanced Micro Devices announced a $10 billion investment in Taiwan's semiconductor ecosystem, focused on accelerating AI infrastructure through advanced packaging and manufacturing partnerships.
The investment targets three areas: scaling advanced packaging capabilities, developing next-generation interconnect technology, and building rack-scale AI systems for deployment starting in 2026. AMD is working with Taiwanese manufacturers ASE and SPIL on the core technical work.
Elevated Fanout Bridge Technology
At the center of the initiative is EFB (Elevated Fanout Bridge) technology, a 2.5D interconnect that increases bandwidth between chips while reducing power consumption. AMD and its partners are qualifying this technology for use in future processors.
PTI achieved a significant milestone by qualifying the industry's first 2.5D panel-based EFB interconnect. This approach enables high-bandwidth connections at scale, which matters for AI systems that need to move large amounts of data between components quickly.
Helios Platform and 2026 Deployment
AMD is integrating EFB technology into its Helios rack-scale platform, which combines 6th Gen EPYC CPUs (codenamed "Venice") with Instinct MI450X GPUs. The company plans multi-gigawatt deployments beginning in the second half of 2026.
Original design manufacturers including Sanmina, Wiwynn, Wistron, and Inventec are building systems around the Helios platform. This arrangement connects silicon design, packaging, and system integration in a single supply chain.
Why Taiwan Matters
Taiwan remains central to global semiconductor manufacturing and advanced packaging. AMD's substantial investment signals the company's reliance on Taiwanese partners for scaling AI infrastructure as demand for compute continues to grow.
The strategy reflects a broader industry shift: AI performance increasingly depends on how efficiently chips connect to each other, not just individual chip performance. Advanced packaging solves that problem at scale.
For IT and development professionals, this investment affects the infrastructure you'll deploy on. Learn more about AI for IT & Development to understand how these hardware advances translate to practical system design.