The increasing size of AI chips is bringing packaging materials and processing equipment to the forefront.
On June 8, Huatai Securities' machinery equipment team, including Yang Yunxiao, wrote in a research report: "The rapid growth in demand for AI computing chips and the existing shortage of packaging materials are accelerating the shift towards new advanced packaging materials such as glass, ceramics, and M8/M9-grade PCBs."
The key part of this statement is the latter half: glass, ceramics, and M8/M9-grade PCBs are all difficult to process. Traditional mechanical drilling is prone to chipping and cracking; wet etching has limited efficiency and profile control; and conventional lasers easily cause thermal damage. This is precisely where the value of ultrafast lasers lies.
"Traditional mechanical drilling, wet etching, and conventional laser processing are inadequate, while ultrafast lasers, with their cold processing characteristics, have become the solution for precision machining."
The firm's calculation framework includes glass interposers, glass substrates, M9 materials, and optical module substrates as potential applications, projecting a long-term market potential exceeding one trillion yuan. However, this potential is not realized linearly; it depends on whether technologies like CoPoS, CoWoP, glass substrates, and M9 PCBs can enter mass production.
The Larger the AI Chip, the Sooner Packaging Materials Hit a Bottleneck
The pressure for advanced packaging first comes from the chips themselves.
In Nvidia's product iterations, the number of chips within a package and the HBM configuration have been continuously increasing. Data shows that the GP100 used 4 HBM stacks with 16GB capacity and integrated 5 chips; the GB100 has reached 8 HBM stacks, 192GB capacity, and integrates 10 chips.
As chips grow larger and package areas increase, challenges with heat dissipation, warpage, and signal transmission become more difficult.
The current mainstream CoWoS path is divided into S, R, and L types. CoWoS-S offers the best performance but at the highest cost, with a maximum package size of about 2700 mm²; CoWoS-R is lower cost, but controlling warpage in large packages is challenging; CoWoS-L strikes a balance between cost and performance, but still has limitations in maximum size, thermal dissipation capability, and reliability.
The next steps in the roadmap primarily look at CoPoS and CoWoP.
CoPoS is about "turning a circle into a square." It changes the packaging carrier from a round wafer to a square panel to improve area utilization and uses glass to replace silicon or organic interposers.
CoWoP is more direct: it eliminates the ABF substrate, directly bonding the silicon interposer to the PCB. This shortens the interconnect path but imposes higher demands on the PCB, including finer line width/spacing, void-free via filling, and low thermal expansion materials.
According to TrendForce and other sources, TSMC plans to establish a CoPoS mass production line at its Chiayi facility, with equipment delivery for the pilot line starting in February 2026. At TSMC's annual shareholder meeting on June 4, Chairman and CEO C.C. Wei mentioned that pilot lines for CoPoS and glass substrates have already been built, with an estimated 2-3 years to reach larger-scale mass production. According to other reports, Nvidia aims to achieve CoWoP mass production on its Rubin Ultra platform.
Material shortages are also pushing the industry forward.
The high-performance ABF substrates used for CoWoS rely on core materials including ABF film and T-Glass, a special low-dielectric glass fabric. T-Glass supply is "almost exclusively from Japan's Nitto Boseki, whose current capacity is fully loaded."
Simultaneously, Nvidia's next-generation Rubin high-end GPUs require high-end ABF substrates for packaging and involve advanced stockpiling, further amplifying supply tightness.
This is one of the reasons glass-based materials are being pushed back into the spotlight.
Glass, Ceramics, and M9 Are Not the Same Layer of Material; They Are Not Simple Substitutes
The market often compares glass, ceramics, and M9 materials together. However, they are not entirely competing on the same level.
Huatai Securities states directly: "Glass substrates are mainly used in the interposer and substrate fields of advanced packaging and do not conflict with materials like ceramic materials/M9 used in PCB applications."
The traditional CoWoS structure, from top to bottom, consists of the chip, interposer, substrate, and PCB layer. Glass is primarily used for interposers and substrates. Ceramic-based and M9 materials are used more for PCB-related materials or high-power thermal dissipation scenarios.
Glass's advantages lie in dimensional stability, surface flatness, low high-frequency loss, and the ability to create TGV (Through Glass Vias). Data shows glass has a dielectric constant of 3.5 to 10, a tunable CTE of 2.7 to 12.4 ppm, and a surface flatness that can be less than 4nm.
The upgrade path for PCBs is towards M8, M9, and even M10.
A higher number indicates lower signal transmission loss, higher speed, and greater stability. The core of M8/M9-grade PCBs lies in lower dielectric constant (Dk), lower dissipation factor (Df), a high-modulus, low-expansion glass fiber system, and ultra-low profile copper foil.
The challenge with M9 is that the material is harder. It introduces quartz fabric (Q-glass) as a reinforcement material. High-purity quartz glass fiber has a Mohs hardness exceeding 7, higher than the 5-6 level of traditional E-glass fiber.
Ceramics focus on thermal dissipation and thermal expansion matching.
Data shows ABF material has a thermal conductivity of 0.8-1.2 W/mK, while ceramic substrates can reach up to 200 W/mK. Aluminum nitride has a thermal conductivity of 170-230 W/mK, silicon nitride 60-90 W/mK, and sintered silicon carbide about 100-250 W/mK.
This explains why all three material types may be in focus simultaneously: glass addresses interposers and substrates, M8/M9 addresses high-speed interconnects, and ceramics addresses high-power thermal dissipation.
Why the "Scalpel": Ultrafast Lasers Minimize Thermal Damage
The key point of ultrafast lasers is not "higher power," but "shorter time."
Huatai Securities defines it: "Ultrafast lasers typically refer to lasers with pulse durations on the order of picoseconds (10⁻¹² s) to femtoseconds (10⁻¹⁵ s)."
In such a short time, the energy arrives and departs quickly. The material does not have time to diffuse the heat before the removal is complete. This mechanism is known as "cold ablation."
This differs from traditional long-pulse lasers.
Traditional lasers are more like "burning away" the material. The heat-affected zone is large, easily causing chipping, micro-cracks, melting, and carbonization.
Ultrafast lasers are more like "stripping away" the material. Through nonlinear effects like multi-photon absorption, they act directly at the electronic level on the material surface, reducing thermal diffusion.
The report's phrasing for this is: "Ultrafast lasers are not a simple parameter upgrade over traditional lasers, but a fundamental change in the processing mechanism."
In practical processing, this difference translates to three results: a smaller heat-affected zone, adjustable via wall taper angles, and more flexible processing shapes.
Analysts note that laser drilling can create micro-vias of any shape, which is difficult for mechanical drilling to achieve.
Three Processes Best Illustrate the Demand: TGV, M9 Micro-vias, and Ceramic Etching
The first is TGV for glass substrates.
The key manufacturing process for glass substrates involves six steps: TGV laser modification, via etching, AOI optical inspection, seed layer deposition, electroplating via filling, and grinding.
Among these, "TGV laser modification is the first and most critical step."
The reason is simple: glass lacks plastic deformation capability. Excessive or uneven energy easily leads to micro-cracks, thermal stress concentration, or internal defects. When the via diameter is below 30 microns, controlling the heat-affected zone becomes even more stringent.
Ultra-short pulse ultrafast lasers can achieve non-thermal modification inside glass, reducing issues with thermal stress cracks and chipping.
The second is M9-grade PCB micro-via processing.
M9 targets ultra-high-speed transmission environments above 1.6 Tbps. To reduce signal loss, it introduces high-purity quartz glass fiber, but this also increases processing difficulty.
Data indicates that the lifespan of traditional mechanical drill bits can plummet to one-fifth that of traditional materials, with via diameter accuracy and positional accuracy also deteriorating.
The problem with CO₂ lasers is an excessively large heat-affected zone, which may cause resin carbonization and glass fiber tearing. Nanosecond UV lasers have low etching efficiency for hard quartz glass fiber and poor via wall quality.
The value of ultrafast lasers lies in their ability to precisely remove high-hardness glass fiber without carbonizing the resin or short-circuiting the copper layer.
The third is fine processing of ceramic substrates.
Ceramic materials like aluminum nitride, silicon nitride, and silicon carbide are hard and have high thermal conductivity. Data shows these materials can have a Mohs hardness of 7-9 and thermal conductivity exceeding 200 W/mK.
Mechanical drilling leads to rapid drill bit wear, rough via walls, and severe chipping. The energy from conventional lasers is quickly diffused, instead forming a heat-affected zone and micro-cracks.
Ultrafast lasers can improve this issue, but there are still boundaries.
Analysts also caution that in high-throughput processing, the thermal accumulation effect from high-frequency pulses can still induce micro-cracks, requiring reasonable control of repetition rate and energy density. The industry is still exploring composite solutions like high-energy ultrafast lasers, mechanical pre-drilling plus ultrafast laser finishing, magnetic field-assisted laser processing, and cryogenic-assisted laser processing.
Where Does the Trillion-Yuan Potential Come From?
The market potential for ultrafast laser equipment primarily stems from four types of potential applications: CoPoS glass interposers, ABF glass substrates, M9 materials, and optical module substrates.
In the calculation, the unit price of the equipment is set at 6 million yuan. Under different penetration rate assumptions, the corresponding potential market sizes are:
10% penetration rate: approximately 10.3 billion yuan.
30% penetration rate: approximately 31.0 billion yuan.
50% penetration rate: approximately 51.6 billion yuan.
80% penetration rate: approximately 82.6 billion yuan.
100% penetration rate: approximately 103.3 billion yuan.
Among these, M9 materials contribute the most. Under the 100% penetration assumption, M9 materials correspond to a demand for 11,574 units; ABF glass substrates correspond to 3,704 units; CoPoS glass interposers correspond to 1,757 units; and optical module substrates correspond to 174 units.
This set of calculations has two underlying premises.
First, that advanced packaging indeed expands from CoWoS to CoPoS and CoWoP. TSMC has planned a CoPoS mass production line and has already built pilot lines for CoPoS and glass substrates, estimating entry into larger-scale mass production in 2-3 years.
Second, that materials like glass-based, M9 PCB, and ceramics are not just laboratory paths but can enter scaled manufacturing. Equipment orders ultimately come from mass production lines, not from technological narratives.
LPKF Holds the High End; Domestic Players Chase Complete Solutions
In the global ultrafast laser equipment landscape, overseas leaders currently hold a first-mover position in the high-end market, while domestic manufacturers are accelerating their catch-up.
Germany's LPKF has an advantage in its LIDE (Laser-Induced Deep Etching) process. This process is used for glass substrate processing, enabling crack-free, high aspect ratio, low thermal damage machining. Its Vitrion S 5000 targets thin glass micro-machining and TGV vias, with a maximum aspect ratio of up to 1:50.
The path for domestic manufacturers is more diversified and closer to downstream process validation.
Han's CNC, focused on PCB-specific equipment, has launched ultrafast laser drilling equipment. Its glass laser drilling machine can achieve a minimum via diameter of 10µm, with an aspect ratio up to 50:1 for mainstream materials.
Han's Laser possesses picosecond UV, picosecond IR, and other light source technologies. Its products cover micro-machining of brittle materials like glass, sapphire, and ceramics, and are also compatible with advanced processes like mSAP and TGV.
DR Laser focuses on glass TGV. Its wafer-level TGV laser micro-via equipment supports different glass materials, with a minimum aperture ≤5µm and an aspect ratio as high as 1:100.
Inno Laser has a product portfolio ranging from nanosecond to femtosecond and from infrared to deep ultraviolet, with laser unit sales exceeding 22,000 units. Its PCB ultrafast drilling equipment has secured its first order.
Win Laser's glass laser drilling machine has a repeat positioning accuracy of ±1µm, can process glass up to 20mm thick, and its glass TGV drilling equipment is in the customer validation stage.
Delong Laser specializes in precision laser micro-processing equipment and core laser units, with self-developed picosecond and femtosecond solid-state lasers covering scenarios like TGV, wafer dicing, and ceramic and glass processing.
Hymson Laser centers on "laser + automation," deploying processes like laser etching and laser-induced processes, with business covering lithium batteries, photovoltaics, consumer electronics, and semiconductors.
The focus of competition is no longer just the parameters of a single machine. Advanced packaging materials are complex, process windows are narrow, and downstream players increasingly need complete solutions encompassing the light source, motion control, process parameters, inspection, and automation. The opportunity for domestic manufacturers also comes from localized delivery and engineering response capabilities.
Risk Lies in Mass Production Timelines, Not the Concept Itself
The logic for ultrafast lasers is clear: the harder the material, the smaller the via, and the less acceptable thermal damage is, the more valuable ultrafast lasers become.
However, there are three categories of risk for equipment volume ramp-up.
First, slower-than-expected development of advanced packaging technologies. If the advancement of paths like CoPoS, CoWoP, and glass substrates lags behind expectations, related equipment demand will also be delayed.
Second, slower-than-expected AI computing investment. The fundamental driver for packaging material upgrades comes from demand for high-end AI chips. If downstream capital expenditure slows, equipment orders will be affected.
Third, risks from international trade friction. Advanced packaging materials and equipment rely on global supply chains. Trade restrictions could impact validation, delivery, and customer expansion schedules.
For the market, ultrafast lasers are not merely a "laser equipment" story. They are the precision tool that the manufacturing segment must acquire following the switch in advanced packaging materials.
Comments