Four glass manufacturers have standardized on either 510x515mm or 515x510mm, driven by existing equipment and display glass supply chains, despite unresolved process route convergence.
Reviewing public documentation from these companies reveals a subtle inconsistency. SCHOTT's glass panel technical specifications list a standard format of 510 x 515 mm; AGC's TGV product specifications state 510x515mm; Nippon Electric Glass (NEG) cited 515x510mm in an official January 2025 press release; and DNP's pilot line in Kuki, Saitama, also specified 510x515mm. The same panel, two notations, both formalized in specifications, with no vendor correcting the other.
In the packaging industry, dimensions are typically the first parameter fixed and the hardest to reverse, as the entire production line is organized around them. Therefore, when a specific millimeter-level dimension begins to repeat across competing vendors, the critical question is what drove its selection.
The primary driver is existing equipment inventory
NEG stated the rationale explicitly in its official press release: 515x510mm is a large panel format already adopted in many semiconductor manufacturing processes, allowing customers to leverage existing equipment and reduce capital investment. Glass substrates are replacing organic substrates currently running on customer lines, and the last thing customers want is to build an entirely new line just to switch materials. Aligning the format with existing panel-level equipment sets a floor for adoption costs.
The second force comes from display glass. According to industry convention, the 510mm tier aligns with the cutting lines for LCD generation panels, meaning Corning and AGC can directly cut this specification from their existing G4/G5 glass blanks, significantly lowering raw material procurement costs compared to custom formats. The semiconductor industry is borrowing the amortized material costs of the panel industry.
0.263m²
Single panel area
2.73x
Relative to 310x310mm
3.7x
Relative to 300mm wafer
These three figures represent the area of a single 510x515mm panel, its area multiple relative to the industry transitional specification of 310x310mm, and its area multiple relative to a 300mm wafer. Area multiples directly correspond to the number of chips that can be arranged per batch.
Same format, four process routes
All four companies are betting on the same form factor, but the methods for drilling holes and filling them with copper vary significantly. NEG is pursuing two parallel tracks: one uses CO2 lasers to directly process glass-ceramic cores (GC Core) for speed and reduced cracking risk; the other employs laser modification followed by wet etching to achieve finer holes. On May 22, 2025, it announced two types of large-format TGV glass core substrates, with the laser-modified variant already in sampling, achieving a hole diameter of 50µm on 0.4mm-thick boards.
AGC is utilizing alkali-free glass EN-A1, with hole diameters ranging from φ20 to 150µm and an aspect ratio of up to 20:1 on 1.0mm-thick boards. DNP is preparing two metallization approaches: a filled type that completely fills holes with copper, and a conformal type that applies a metal layer only to the hole walls, targeting an aspect ratio greater than 1:15.
Thickness standards are also misaligned. Schott’s panels start at 0.5mm, while NEG’s GC Core is 1.0mm; Korean firm Philoptics displayed samples up to 2.0mm thick at KPCA SHOW 2026 in Incheon in September. A 2mm thickness allows for the creation of blind cavities to embed multilayer ceramic capacitors within the board, a fundamentally different packaging design approach compared to 0.4mm thin boards.
💡 Tip
Even with the same 510×515mm format, hole diameters can range from 50µm to 150µm; metallization can be full copper fill or hole-wall only; and board thickness can vary between 0.4mm and 2.0mm. Uniformity in format ensures compatibility of carriers and equipment, but packaging designs remain non-universal.
Dimensions are converging, timelines are diverging
Lining up the timelines reveals a different picture. DNP’s pilot line at its Kuki, Saitama plant will begin phased production in December 2025, with external sampling starting in early 2026 and mass production targeted for fiscal year 2028. NEG’s glass-ceramic core for CO2 laser processing also targets mass production around 2028. The schedules of these two Japanese manufacturers are closely aligned.
On another front, LG Innotek CEO Moon Hee-soo announced at CES in January 2026 that the mass production timeline for glass substrates has been pushed back from 2028 to 2030, citing that current demand levels do not justify the capital expenditure. Product development is complete; the bottleneck is the investment pace. In the same business, one company is expanding its pilot line while the other is delaying its timeline.
Domestic efforts are unfolding in parallel. In a September 2026 interview, Vog Optoelectronics Chairman Zhang Fujia stated that the company has invested nearly 1 billion yuan to build a TGV production line with a monthly capacity of 10,000 wafers, utilizing the 510x515mm panel-level format. However, the semiconductor packaging application remains in the design verification stage, with Micro LED and optical modules leading initial volume production. On September 22, BOE Chairman Chen Yanshun announced at IPC2026 that the company is preparing to invest in a mass production line for semiconductor glass substrate package substrates. Its panel-level pilot line achieved full-line commissioning in the first half of 2026, with a designed capacity of 1,000 units per month. Samples for 9-2-9 and 20-layer configurations have been completed and delivered, with initial mass production planned for early 2027.
A falsifiable judgment on this matter: by the end of 2027, 510x515mm will become the default format for new glass substrate pilot lines in mainland China, Japan, and South Korea, but this consensus remains limited to physical dimensions. The true differentiator determining who achieves scale first is the combined certification of glass formulation, drilling processes, and metallization; board size itself does not constitute a barrier. The most likely reason this judgment could be wrong is if new lines selecting 600x600mm or larger formats appear before the end of 2027, indicating that the size consensus never materialized. Another reference point: LG’s decision to push mass production to 2030 better reflects the current realistic pace of this business than four suppliers listing the same number on their specification sheets.
Four manufacturers have written the same number on the same line; who achieves yield first will likely become clear within a year. One thing is unlikely to change: the line that first achieves stable yield will transform the 510x515mm format from a product specification into a de facto standard.