IN Brief:
- SEMI and Global Net Corp. forecast a 67.2% CAGR for glass-core substrates from 2028 to 2040.
- The technology is being evaluated for AI, HPC, advanced processors, co-packaged optics, and image sensors.
- Larger package sizes, finer interconnects, and dimensional stability demands are pushing substrate development beyond conventional materials.
SEMI and Global Net Corp. have released a market report forecasting rapid long-term growth for glass-core substrates as AI and high-performance computing packages place greater demands on advanced semiconductor packaging.
The report examines glass-core substrates as a possible next-generation platform for larger and more complex semiconductor packages. Based on an average of positive, base, and negative market-development scenarios, it projects a compound annual growth rate of 67.2% from 2028 to 2040.
Limited-volume production could begin around 2028 in selected high-performance applications, with wider adoption developing as qualification, process control, supply-chain readiness, and standards mature. The report identifies potential use cases in AI, HPC, advanced processors, co-packaged optics, and image sensors.
Glass-core substrates are being evaluated because they can support larger package sizes, finer interconnects, and improved dimensional stability compared with conventional package-substrate materials. Those properties are becoming more valuable as advanced processors move toward heterogeneous integration, chiplets, wider interfaces, and memory-adjacent architectures.
That packaging pressure is already visible across the sector. TPK and ASE’s planned through-glass via packaging pilot line for 3Q26 points to the same shift from laboratory exploration toward process learning, equipment selection, and yield validation.
Advanced packaging has moved from a supporting process into one of the main determinants of semiconductor system performance. Device scaling alone can no longer carry the full performance roadmap, particularly in AI accelerators and high-density compute systems where die-to-die bandwidth, power delivery, thermal dissipation, routing density, and package warpage have to be solved together.
Organic substrates and silicon interposers will remain central to high-end packaging, but both face rising engineering pressure as package dimensions expand. Large AI accelerator packages and chiplet-based processors require substrate technologies that can maintain flatness, support fine routing, and limit mechanical distortion across more demanding assembly processes.
Glass offers a promising route, although the commercial path remains difficult. Through-glass vias, metallisation, inspection, panel handling, yield control, and integration with existing assembly and test flows all need semiconductor-grade repeatability. Any new substrate platform also has to compete with incumbent ecosystems that already have production knowledge, supplier relationships, and embedded design rules.
The report places glass-core substrates in an exploratory but increasingly serious phase of development. AI and HPC are forcing package-level innovation to accelerate, and substrate materials are now central to that work. If glass can move from pilot lines into stable manufacturing flows, it could become one of the enabling technologies behind the next generation of high-density compute packages.
