The true meaning of "inaugural year" is that the technological path is set, industry giants are beginning to invest in mass production lines, but large-scale commercialization still requires 3 to 5 years.
On June 5th, shares of Boe Technology Group Co.,Ltd. (ASX: 000725) hit a limit-up during the session before pulling back, ultimately closing up over 4%. This marked the second consecutive day the stock challenged the daily gain limit. The company's latest total market capitalization approached 240 billion yuan, with trading volume reaching a massive 35.7 billion yuan, setting a new record high.
The catalyst likely stems from news over a month ago. On May 20th, BOE signed a three-year memorandum of cooperation with global glass giant Corning. The two parties will jointly advance the application of glass substrates in the field of semiconductor packaging. The true significance of this partnership lies not in display panels, but in chips. A company primarily known for display panels is stepping into the semiconductor packaging materials arena. The capital market has cast its vote of approval for this cross-sector move with two limit-up rallies.
A Material Revolution Catalyzed by Limit-Ups
The immediate trigger for BOE's limit-ups is the glass substrate concept. However, glass substrates are not merely an extension of the panel industry; they represent a material revolution in semiconductor packaging. Traditional chip packaging relies on two types of materials: organic substrates and silicon interposers. Organic substrates are low-cost but prone to warping at high temperatures and suffer from significant high-frequency signal loss. Silicon interposers offer good performance but are expensive and have low yield rates for large sizes. With the explosive growth in AI chip power consumption, both paths have reached their limits.
Glass substrates replace organic substrates and silicon interposers with specialty glass. Leveraging four major advantages—low thermal expansion, low dielectric loss, high flatness, and suitability for large sizes—they have become a crucial alternative solution for AI chip packaging. According to Intel's technology roadmap, glass substrates can achieve over a 10x improvement in interconnect density, with signal transmission and power consumption performance significantly superior to organic substrates. In an industry report released on May 17th, Western Securities assigned an "overweight" rating to the glass substrate industry. The core logic is that 2026 marks the official commercial inaugural year for this new material, which has extended from the display field into semiconductor packaging.
Collective Bets from Global Giants
The glass substrate story is not a solo performance by one company but a collective wager by global semiconductor giants. Intel has been involved the earliest and with the greatest commitment. It officially released glass substrate samples in September 2023, has established a pilot line in Chandler, Arizona, and plans to convert its Rio Rancho, New Mexico facility into the world's first glass substrate mass production base. Intel previously entered a cooperation framework with Brookfield Infrastructure, where the latter invested approximately $13 billion for a 49% stake in the manufacturing expansion at Intel's Arizona campus. The initial investment for the Rio Rancho base exceeds $1.5 billion, with yield rates already surpassing 95%.
TSMC is following closely. Its Chip-on-Polymer-on-Substrate (CoPoS) pilot line is scheduled to start in 2026, with mass production planned for 2028, with NVIDIA as a primary customer. Samsung Electro-Mechanics' pilot line at its Sejong plant has achieved a breakthrough in Through-Glass Via (TGV) aspect ratio of 10:1. It is not only supplying Apple with samples of its "Baltra" AI server chip but also pushing Samsung Electronics to test glass substrates for HBM4 packaging. SKC/Absolics has built the world's first mass-production-scale factory in Georgia, USA, and has already provided mass-production-grade samples to AMD.
Corning occupies the most upstream and technologically challenging segment of the entire industry chain, holding an estimated 70% share of the global semiconductor glass wafer market. It is the exclusive supplier to NVIDIA, AMD, and TSMC. The essence of BOE's cooperation with Corning is that the panel leader aims to leverage Corning's glass wafer technology to enter the semiconductor packaging track.
A Supply-Demand Gap in a Trillion-Yuan Market
The core driver for glass substrates moving from the laboratory to industrialization is the rigid demand-supply gap for packaging materials driven by AI chips. According to Omdia data, the overall global glass substrate market size in 2026 is approximately $18.6 billion (including sub-markets like displays, semiconductors, and storage), with the potential to exceed $32 billion by 2030. Among these, glass substrates for semiconductor packaging are the fastest-growing segment, with multiple institutions forecasting a compound annual growth rate exceeding 30%. The TGV laser equipment market is expected to reach around 3 billion yuan in 2026, with year-on-year growth exceeding 50%. Orders for leading overseas equipment suppliers are already booked through the end of 2027.
However, the supply side is severely lagging. Industry analysis suggests a supply-demand gap of about 40% for semiconductor packaging glass substrates by 2026. The reason lies in the extremely high technical barriers of TGV processing—micro-vias need to be on the 3-micron scale with an aspect ratio of 150:1, making yield control an industry-wide challenge. While Intel's yield exceeds 95%, this represents only the top global standard, with the industry average yield far below this figure.
Localization rates are also a concern. Corning monopolizes over 70% of the glass wafer market. Domestic TGV equipment and glass core substrates are still in the initial stages of development from zero to one, indicating vast potential for import substitution. This is a nearly untapped market but also a technological wall difficult to surmount in the short term.
The Confidence and Chasm of Cross-Sector Expansion
BOE's confidence in pursuing glass substrates stems from its experience in panel-level mass manufacturing. In the display panel field, the company has accumulated years of expertise in glass processing, precision etching, and yield control, capabilities that could potentially be partially transferred to the semiconductor domain.
Yet, reality is harsher than imagination. In 2024, BOE invested only 993 million yuan to build a pilot line for glass substrate packaging carriers. According to the company's institutional research record dated June 3rd, as of that date, this business had not yet achieved batch production or mass production revenue. The pilot line's yield had not reached mass production levels, with significant uncertainty regarding when it would. The company has only sent samples to some domestic customers, with some clients just passing the concept validation stage and entering technical testing.
Display-grade and semiconductor-grade glass substrates are two distinct tracks. Semiconductor packaging demands for thermal expansion coefficient, surface roughness, and micro-via precision are an order of magnitude higher than for display panels. BOE's 2025 annual report shows R&D expenses of 13.983 billion yuan, accounting for 6.83% of revenue. These R&D investments are tilting towards semiconductor materials, but technological leaps require time.
A more immediate issue is that glass substrates are likely to contribute almost zero to BOE's 2026 financial performance. The 246 billion yuan market capitalization and the 24.4 billion yuan record trading volume essentially represent the market pricing a distant narrative in advance. The market is betting that BOE can replicate its success in the LCD field—from catching up to surpassing. However, the players in semiconductor packaging are Intel, TSMC, and Samsung, with competition intensity far exceeding that of the panel industry.
The Most Underpriced Upstream Material
Within the entire glass substrate industry chain, the segment with the thickest profits, highest technical barriers, and arguably the least fully priced by the capital market is the upstream materials sector. TGV processing requires two core technologies: glass wafers and laser drilling equipment. Corning dominates the glass wafer field, while overseas manufacturers like Germany's LPKF monopolize the laser drilling equipment market. The question for domestic substitution is not "if" but "how long it will take to achieve a breakthrough."
However, change is underway. While the TGV laser equipment market is projected to be around 3 billion yuan in 2026—not a massive total—its year-on-year growth exceeds 50%, and domestic equipment is beginning to achieve breakthroughs from zero to one. The investment logic in this field shares similarities with previously discussed areas like MLCCs: a small market can ignite a large industry chain. The global glass substrate market is estimated at $18.6 billion in 2026, but the AI/HBM semiconductor packaging segment is growing at a CAGR exceeding 30%, acting as the fastest-growing engine for this market. Capital's valuation of this track has only just begun.
Beyond the Inaugural Year of Mass Production
2026 is regarded as the inaugural year for glass substrate mass production. But the true meaning of this "inaugural year" is that the technological path is set, industry giants are beginning to invest in mass production lines, but large-scale commercialization still requires 3 to 5 years. Intel is the most aggressive player, with its initial investment exceeding $1.5 billion demonstrating its resolve. However, it's worth noting that Intel previously faced rumors of "halting" its glass substrate business, which were later clarified as misunderstandings. This indicates that even for a giant like Intel, the commercial realization of glass substrates faces uncertainties. TSMC's CoPoS won't reach mass production until 2028, and Samsung's HBM4 glass substrate solution is slated for 2027—these timelines themselves illustrate that there are no shortcuts on this path.
For BOE, the cooperation with Corning is an entry ticket, not the finish line. Panel-level mass manufacturing capabilities provide a cost advantage, but semiconductor packaging customer qualification cycles can last 12 to 18 months, and wafer fabs have far stricter requirements for material stability than panel manufacturers. The distance from "able to manufacture" to "able to sell" to "able to profit" is vast.
Limit-up rallies related to glass substrates will likely recur because the narrative is compelling—AI computing power is a rigid demand, it's a material revolution, there's a trillion-yuan space for import substitution, and industry giants are collectively entering the field. However, what will truly determine the height of this material revolution is not the limit-up boards, but every percentage point improvement in TGV processing yield, every process breakthrough in domestic equipment moving from zero to one, and every page of the calendar turned on the path towards domestic substitution of materials like Corning's glass wafers.
Comments