Glass substrates are driving advanced packaging into a new round of competition: what does SEMI’s latest report mean?

Glass Substrates Are Driving Advanced Packaging Into a New Round of Competition: What Does the Latest SEMI Report Mean?

The new report released by SEMI and Global Net Corp. has pushed “glass core substrates” further from the lab and industry discussions into the center of attention in the semiconductor supply chain. The report makes it clear that this technology is attracting more industry investment, mainly driven by demand from AI and high-performance computing (HPC) for larger and more complex packages; its market forecast shows that glass core substrates may begin entering some high-performance applications after 2028, and long-term adoption will depend on the maturity of materials, processes, yields, equipment, and the supply chain.

The significance of this news is not that “glass substrates have already won,” but that it sends a clearer industry signal: as the marginal gains from process scaling slow down, competition in system performance is shifting toward packaging. For GPU, AI accelerator, advanced processor, and co-packaged optics (CPO) vendors, packaging is no longer just a back-end cost, but a key variable determining bandwidth, power consumption, size, and thermal design. In other words, the focus of the next round of semiconductor competition is shifting from transistor density in a single chip to integrated “chip–package–system” capability.

Background: Why Glass Substrates Have Suddenly Drawn Attention

Traditional high-end packaging has long relied on organic substrates, especially ABF substrates, which have already formed a mature supply chain in server GPUs, AI ASICs, and high-end processors. However, as package sizes continue to grow and I/O density keeps rising, traditional materials are beginning to face greater pressure in warpage control, dimensional stability, and finer-line capabilities. SEMI noted in the report that glass core substrates are seen as a potential solution precisely because they may support larger package sizes, finer interconnects, and better dimensional stability.

This means glass substrates are not an isolated material replacement, but a node in the evolution of advanced packaging. They, together with 2.5D interposers, HBM stacking, hybrid bonding, Chiplet architectures, and CPO, form a combined technology stack. For AI chips, what is truly scarce is not computing power itself, but the ability to “organize” that computing power in a high-bandwidth, low-latency, low-power way. Packaging upgrades are, in effect, replacing part of the system interconnect challenge.

Industry Chain Analysis

Upstream: Who Will Benefit First from Materials and Equipment

If glass core substrates move from R&D to mass production, the first beneficiaries upstream will not be end-chip makers, but the materials and process equipment ecosystem. Glass materials, microvia processing, surface treatment, copper deposition, electroplating, cutting, cleaning, inspection, and metrology all require new process windows. The SEMI report specifically mentions that companies and research institutions are advancing related development simultaneously in Asia, North America, and Europe, which means glass substrates are not a closed innovation in a single region, but a cross-border competition in materials and equipment.

CONTEXT_AFTER: Equipment-side concerns will also change.The focus on the equipment side will also change. ASML is not a direct equipment beneficiary of glass substrates, but the upgrade of advanced packaging routes will drive overall back-end equipment investment; vendors such as Applied Materials, Lam Research, and KLA may see more demand in deposition, etching, inspection, and metrology. For packaging equipment suppliers, a new material system means more demanding requirements for flatness control, alignment accuracy, and defect inspection. Whoever can first solve large-size glass processing, warpage control, and yield issues will be closer to large-scale orders.

Midstream: The role of advanced packaging foundries will become more critical

Packaging houses such as ASE and Amkor, along with the Taiwanese and Japanese OSAT and advanced packaging ecosystems, will play a bridging role in this round of technological iteration. The reason is simple: glass substrates are not a material that can be “automatically adapted after chip design is complete”; they require packaging foundries to redefine process capabilities, equipment configurations, and quality control standards.

This will also strengthen the strategic position of advanced packaging across semiconductor manufacturing. In the past, fabs held the leverage in advanced process nodes; in the future, as chiplets, multi-die integration, and heterogeneous integration expand, advanced packaging capacity itself may become a scarce resource. For TSMC, advanced packaging platforms such as CoWoS and SoIC have already become one of the key bottlenecks in the AI supply chain; for Samsung Foundry and Intel Foundry, if they want to win more customers in the AI and HPC era, packaging capabilities are also a key bargaining chip. If glass substrates mature, they will further amplify the trend that “packaging determines system competitiveness.”

Downstream: Who will adopt first

Based on the application directions given in the SE​MI report, the most likely early adopters of glass core substrates are high-end applications such as AI, HPC, advanced processors, CPO, and image sensors. This does not mean consumer electronics will immediately switch material systems. A more realistic path is to pilot them first in high-value, low-margin-for-error, performance-first scenarios, and then gradually expand.

These applications share common traits: high packaging value per unit, strong performance requirements, and willingness to pay a premium for reliability and supply security. As a result, NVIDIA, AMD, the Google TPU ecosystem, Amazon Trainium, and some AI ASIC customers will benefit from packaging upgrades earlier than smartphone SoC vendors. For Apple Silicon, Qualcomm, MediaTek, and Broadcom, the short-term significance of glass substrates lies more in observing whether the evolution of high-end packaging materials will gradually move down into broader SoC and connectivity chip applications.

Technology ImpactThe core technical value of glass core substrates lies in their potential to alleviate the physical constraints of traditional organic substrates in large-size, high-density interconnect scenarios. The three directions emphasized in the SEMI report are especially critical: larger packages, finer interconnects, and higher dimensional stability. For AI chips, this is the foundation for high-bandwidth interconnects and power consumption control.

However, the technical barriers are equally evident. First, the material itself is only the starting point; the real challenge lies in manufacturing consistency at large sizes and processing yield. Second, glass is a brittle material, making handling, cutting, drilling, and interlayer reliability more challenging than with traditional organic materials. Third, the entire process chain must be compatible with the existing packaging ecosystem; otherwise, even if performance is better, it will be difficult to scale due to cost and long adoption cycles.

Therefore, glass substrates are more like a “candidate for the next-generation high-end packaging platform” rather than a solution that will fully replace ABF substrates in the short term. Their commercialization path is more likely to be layered: first introduced in a small number of high-end AI/HPC packages, then gradually expanded as equipment, materials, and process standards mature.

Competitive Landscape

This competition is not a contest between individual companies, but between industrial platform ecosystems. TSMC relies on the integration of advanced process and advanced packaging to bind customers, creating synergy between “process + packaging”; Samsung Foundry hopes to improve customer lock-in through more complete foundry and back-end integration; Intel Foundry needs to prove its execution capability in advanced packaging and system-level integration. If glass substrates enter mass production, they will further amplify differentiation among packaging platforms.

For chip design companies, the focus of competition is also changing. NVIDIA’s advantage comes not only from its GPU architecture, but also from its ability to organize the entire AI system; AMD needs to continue strengthening the synergy among CPU, GPU, and packaging; Broadcom and the custom ASIC projects of multiple cloud vendors are increasingly dependent on advanced packaging delivery capabilities. In other words, future “chip competition” often takes the form of “who can secure better packaging resources, and who can deliver system performance faster.”

Regional Implications

United States

The United States holds a strong position in AI chips, system architecture, equipment, and some materials R&D. Demand related to glass substrates will further strengthen the U.S. influence in defining high-end computing chips and advanced packaging, but mass production capabilities still depend heavily on the Asian supply chain.

Taiwan

Taiwan remains the core hub of advanced packaging. Whether it is TSMC’s advanced packaging platform or the packaging/test and materials ecosystem, Taiwan occupies a critical position in the AI supply chain. If the adoption of glass substrates accelerates, Taiwanese companies are expected to be the first to take on the complex process tasks of pilot production and scaled deployment.

South Korea

South Korea has strengths in memory, advanced logic, and certain materials and equipment. For the synergy between HBM and AI packaging, the storage and packaging support capabilities of Korean manufacturers remain a key link in the industrial chain.

Japan

Japan continues to have a deep foundation in materials, chemicals, substrates, and highly reliable manufacturing.### Japan

Japan still has a deep foundation in materials, chemicals, substrates, and highly reliable manufacturing. The maturity of glass materials, precision processing, and the high-end materials supply chain will determine where Japanese companies stand in this new track.

Europe and Southeast Asia

Europe places more emphasis on equipment, materials, and R&D collaboration, while Southeast Asia continues to play the role of taking on outsourced assembly and testing, as well as some manufacturing transfer. As packaging complexity rises, whether Southeast Asia can move from a cost-driven role to a destination for high-end packaging will be a key point to watch.

Supply Chain Impact

From a supply chain perspective, the significance of glass substrates lies in shifting one of the key bottlenecks in advanced packaging partly from the “chip front end” to the “packaging back end.” This will bring three outcomes:

1. Capital expenditure for high-end packaging materials and equipment will increase; 2. Packaging yield and delivery cycles will become more important; 3. Regional concentration in the supply chain may, in the short term, actually increase.

At the same time, export controls and geopolitical factors will also affect the pace of adoption. AI chips and advanced packaging are already at the core of U.S.-China technological competition, and any restrictions involving key materials, equipment, and the EDA/IP ecosystem will affect the ramp-up of glass substrate-related capacity. For Chinese manufacturers, the more realistic near-term focus is on material substitution, accumulation of packaging process know-how, and collaboration with domestic equipment suppliers.

Market Watch

SEMI’s forecast of a 67.2% CAGR is a typical early-stage market signal: a small base, fast growth, but not necessarily certain volume expansion. Investors should pay closer attention to three things: first, whether pilot production begins around 2028; second, which customers adopt it first; and third, whether yield and cost approach the threshold for large-scale commercial use.

The reason capital markets are paying attention to advanced packaging is not only because it is “new material,” but because it may redefine the distribution of profits in semiconductor manufacturing. In the past, value was more concentrated in front-end processes; in the future, the bargaining power of high-end packaging, materials, and equipment may continue to rise.

Long-Term Outlook

Within 3 years, glass core substrates will still be in the stages of validation, pilot production, and limited high-end adoption, mainly around AI/HPC and a small number of highly complex applications.

Within 5 years, if yield, reliability, and supply chain coordination achieve breakthroughs, it may secure stable orders on some high-end packaging platforms, but it will still be difficult to fully replace traditional substrates.

Within 10 years, the most likely outcome is not that a single material wins out, but that an advanced packaging ecosystem with multiple materials coexists: ABF, glass substrates, hybrid bonding, and more complex Chiplet architectures will develop in parallel, layered by application scenario.

ConclusionThe most important industry judgment in this SEMI report is not “when glass core substrates will reach large-scale production,” but that it confirms a trend: semiconductor competition in the AI era is shifting from the transistor level to the packaging system level. In the future, what will determine a chip company’s competitiveness is not only advanced process nodes, but also who can master more advanced packaging platforms, material systems, and supply chain coordination capabilities.

For the semiconductor industry chain, glass substrates represent an opportunity for “repricing packaging”; for the capital markets, they represent a new growth curve in advanced packaging, materials, and equipment; for national and regional competition, they mean that the center of gravity of advanced manufacturing is further concentrating in regions with a complete packaging ecosystem.