Driven by the explosive demand for AI computing power, large-die packaging is sparking a fundamental paradigm shift across the semiconductor industry. As market leaders push advanced packaging beyond 2.5D wafer-level into panel-level architectures and 3D stacking, second-tier foundries are actively following suit by deploying key technologies such as TSV and Wafer-to-Wafer bonding. With over ten new advanced packaging facilities already planned by foundry providers worldwide, the industry has officially entered the Foundry 2.0 era. Looking ahead to 2027, advanced packaging evolves from an auxiliary service into a primary competitive differentiator, fundamentally reshaping global market dynamics, value distribution, and competitive strategies across the foundry landscape.

Driven by a massive, AI-fueled capital expenditure cycle, WFE and SPE companies are capitalizing on critical technological inflections—such as High-NA EUV, Gate-All-Around architectures, and cryogenic etching—that inherently increase equipment intensity per wafer. At the same time, the industry faces the complex geopolitical reality of China acting as both its largest customer and its most aggressive structural competitor amid stringent export controls.

This presentation focuses on the latest outlook for the global AI server market and key developments across the AI server supply chain. It will examine how supply chain participants are adapting their strategies to address geopolitical uncertainties and the emergence of a G2 market structure, driven separately by North American AI suppliers and Chinese AI solution providers. From a global AI market perspective, in addition to NVIDIA and AMD continuing to advance integrated AI server rack platforms, another key trend to watch is the accelerating development of in-house AI chips by major CSPs, such as Google and AWS, as well as leading Chinese technology companies. Based on the 2026 AI market outlook, TrendForce will also provide in-depth analysis and key insights into the opportunities and major trends expected to shape the AI market in 2027.

This presentation focuses on the data center deployment of North American and Chinese hyperscalers and the market dynamics. As data center build-out is no longer determined by industry factors alone, but increasingly by power availability, grid interconnection, and permitting regimes, TrendForce adopts a combined top-down and bottom-up approach, mapping power supply and regulatory constraints at the jurisdiction level, while tracking deployment campus by campus at the facility level. It will examine how North American hyperscalers are concentrating capacity into gigawatt-scale AI training campuses across a small number of states, and how Chinese hyperscalers, operating under a different set of constraints, are pursuing a distinct deployment path. Based on this framework, TrendForce will also provide in-depth analysis and key insights into the opportunities and major trends expected to shape hyperscaler data center capacity in 2027 and beyond.