AI Chip HBM Integration to Drive 29% ABF Substrate Shortage by 2028; $19.1B Expansion Unlikely to Offer Short-Term Relief

August 2, 2026

Unknown Taiwan

Overview

The rapid integration of High Bandwidth Memory (HBM) into AI chips is set to drive a 39% Compound Annual Growth Rate (CAGR) in ABF substrate demand between 2025 and 2028. This surge is projected to create a supply deficit by 2027, potentially reaching a critical 29% gap by 2028. While leading manufacturers have committed $19.1 billion to capacity expansion, these investments are unlikely to offer immediate relief, threatening to bottleneck next-generation AI and high-performance computing development and pushing up prices for key suppliers.

In Depth

Background

As the traditional scaling benefits of Moore’s Law diminish, advanced packaging technologies have emerged as a critical frontier for enhancing semiconductor performance. In the rapidly evolving sectors of Artificial Intelligence (AI) and High-Performance Computing (HPC), the proliferation of chiplet architectures and heterogeneous integration necessitates increasingly complex and larger packages. Ajinomoto Build-up Film (ABF) substrates are indispensable components in these advanced systems, serving as the foundation for flip-chip ball grid array (FCBGA) packages used in high-performance CPUs, GPUs, and AI accelerators. While ABF substrate shortages have occurred during previous industry booms, the current AI revolution is driving demand at an unprecedented scale and pace, profoundly impacting the entire semiconductor supply chain.

Key Findings

The demand for ABF substrates is projected to surge dramatically, with a Compound Annual Growth Rate (CAGR) of 39% from 2025 to 2028. This accelerated growth is primarily driven by the intensified integration of High Bandwidth Memory (HBM) into AI chips. The trend towards integrating more HBM stacks, coupled with the widespread adoption of 2.5D/3D packaging techniques, necessitates significantly larger and higher-layer-count ABF substrates. For example, AI server chips frequently incorporate 8 to 12 HBMs per GPU, escalating the physical size and complexity requirements for their underlying ABF substrates.

This rapid demand increase is expected to lead to a significant supply deficit, commencing as early as 2027 and potentially expanding to a critical 29% gap by 2028. In response to this robust demand, leading global substrate manufacturers, including Unimicron, Nan Ya PCB, and Kinsus, have initiated substantial expansion plans totaling $19.1 billion. However, due to lengthy lead times for manufacturing equipment installation and the subsequent production ramp-up, these significant investments are not expected to fully alleviate the imminent short-term supply crunch.

The looming ABF substrate shortage poses a significant bottleneck for semiconductor manufacturers aiming to bring next-generation AI and HPC products to market, directly impacting AI chip production schedules and potentially limiting the overall growth rate of the AI industry. While presenting a considerable business opportunity for ABF substrate manufacturers, this situation also underscores the urgent need for manufacturing technology innovation and supply chain strengthening. Long-term solutions will require not only continuous capacity expansion but also advancements in material science and the development of alternative substrate technologies. This supply deficit is poised to remain a critical concern for the semiconductor industry for the foreseeable future.

Source: https://finance.biggo.com/news/4978b0ef-fc31-4d7d-9de1-c14abbafc97b

Get our weekly technology intelligence — free

Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.

Subscribe Free — Weekly Tech Intelligence

By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.