Samsung is not replacing silicon in AI processors with glass. Samsung Electro-Mechanics is developing glass-core package substrates: layers that help connect processor dies, memory and other components to the system board. The change, if it reaches production, would affect how some advanced chips are packaged—not what their transistors are made from.
What “glass replacing silicon” actually means
The phrase blurs together several different parts of a semiconductor package. An AI accelerator’s silicon die contains its transistors and performs computation. A package substrate and, in some designs, a silicon interposer provide dense connections among the processor, memory and the rest of the system.
Samsung Electro-Mechanics—not Samsung Electronics’ wafer-fabrication business—is developing a glass-core substrate for this packaging layer. Its package-substrate overview describes substrates as the high-density circuit layer that carries signals between a semiconductor and the main board (Samsung Electro-Mechanics).
- Silicon die: The processor or memory chip itself. Glass is not replacing its silicon transistors.
- Organic package substrate: A multilayer, resin-based circuit substrate. A glass core could replace the conventional core in some high-end packages.
- Silicon interposer: A silicon layer used in some 2.5D packages to connect a processor with high-bandwidth memory or chiplets. Glass could eventually replace or supplement it in some designs.
- Glass carrier: A temporary support plate used during processes such as wafer thinning or fan-out packaging. It is not the same thing as a permanent glass-core substrate.
“Glass-core substrate” or “glass-based advanced packaging” is therefore more accurate than “glass chip.”
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minute#1 Best Overall
- NVIDIA Volta GV100 Architecture — 4,608 CUDA Cores, 640 1st-Gen Tensor Cores delivering 14 TFLOPS FP32 and 112 TFLOPS deep learning performance for AI training, inference, HPC, and scientific computing workloads
- 32GB HBM2 ECC Memory — 900 GB/s Bandwidth — High-bandwidth memory on a 4096-bit bus with ECC error correction provides the memory capacity and throughput required for the largest AI models, simulations, and datasets
- PCIe 3.0 x16 Interface — 250W TDP — Standard PCIe Gen3 connectivity with passive cooling designed for enterprise rack server deployment in HPE ProLiant, Dell PowerEdge, and Supermicro platforms with adequate chassis airflow
- NVLink — Scale to 96GB Unified Memory — Connect two V100 GPUs via NVLink at 300 GB/s bi-directional bandwidth to scale GPU memory from 32GB to 96GB for larger AI training and HPC workloads
- Multi-Precision Computing — Supports FP64 (7 TFLOPS), FP32 (14 TFLOPS), FP16 (112 TFLOPS) and INT8 precision modes for flexible deployment across training, inference, and scientific simulation workloads
Why AI packages are getting harder to build
AI systems increasingly combine large accelerators, chiplets and multiple high-bandwidth memory (HBM) stacks. Those components need many fast electrical connections, substantial power delivery and a large package footprint. The challenge is not only fitting more transistors onto a die: the package must route signals and power among a growing number of components while staying aligned and reliable.
Samsung Electro-Mechanics has described its AI/server flip-chip ball-grid-array (FCBGA) products as having larger areas and more layers than ordinary package substrates. As packages grow, different materials’ thermal expansion can cause bending, or warpage, complicating alignment and assembly. Samsung’s 2024 demonstration highlighted this large-area packaging challenge and the substrate’s role in connecting components (Samsung Electro-Mechanics, KPCA 2024).
What a glass-core substrate could improve
Flatness and dimensional stability
Glass is highly flat and dimensionally stable, qualities that could help keep large package structures aligned during manufacturing and thermal cycling. Samsung says its glass-core design improves warpage control and signal performance; Intel also identifies mechanical and thermal stability as reasons to pursue glass cores (Intel’s glass-core substrate brief).
Rank #2
- High-Performance AI Processing: The MX3 is designed to handle the most demanding AI computer vision workloads, delivering exceptional performance and efficiency.
- Flexible Integration: The MX3 can be easily integrated into your existing systems via its M.2 M-key form factor and support for Linux operating systems.
- Energy Efficient: The MX3 is designed to provide high performance while minimizing power consumption.
- Comprehensive Software Development Kit (SDK): The MX3 is supported by a comprehensive SDK that simplifies development and deployment.
- Hardware compatability: The MX3 is compatible with the PCI-SIG M.2 M-key 2280 Specification. It can be used with the Raspberry Pi 5 with a M-key 2280 HAT.
Panel-scale manufacturing
Glass can be processed in large rectangular panels, which may make better use of material as package footprints expand. Corning lists carrier formats of approximately 515 × 510 mm and 600 × 600 mm, while AGC describes panel-format production for through-glass-via (TGV) substrates. These examples concern glass capabilities and carriers; they do not establish that Samsung’s permanent substrate is already made at those formats (Corning; AGC).
Fine routing and potential signal benefits
A smooth, stable glass surface can support fine redistribution layers, while TGVs can provide vertical connections through the glass. AGC describes TGV structures, fine pitches, cavities and high-aspect-ratio features for advanced packaging applications. Glass may also offer favorable dielectric properties for high-frequency signals, but the result depends on the complete design and operating conditions; there is no single performance gain that applies to every glass package.
A thinner showcased design
At KPCA 2025, Samsung Electro-Mechanics said its demonstrated glass-core substrate was about 40% thinner than conventional substrates. That is a company-reported comparison for the design it showcased, not a general result for glass substrates or proof of a 40% improvement in chip speed or power use (Samsung Electro-Mechanics, KPCA 2025).
Rank #3
- ✅Powered by 26 Tera-Operations Per Second (TOPS) Hailo-8 AI Processor. 2.5W typical power consumption
- ✅Scalable, enabling simultaneous processing of multi-streams & multi-models
- ✅Enabling real-time, low latency and high-efficiency AI inferencing on the edge devices
- ✅Supports TensorFlow, TensorFlow Lite, ONNX, Keras, Pytorch frameworks
- ✅Supports Linux and Windows. Supports the temperature range of -40°C to 85°C
Samsung’s progress—and what remains unproven
Samsung Electro-Mechanics publicly demonstrated a glass substrate in 2024, then described a pilot line and a mass-production target from 2027 onward in a January 2025 announcement. In September 2025 it showed glass-core substrates alongside AI/server FCBGA, 2.1D packaging and co-package technologies. In November 2025, it signed a memorandum of understanding with Sumitomo Chemical Group and Dongwoo Fine-Chem to explore a joint venture. The MOU was not a completed joint venture or a production contract.
| Date | Announcement | What it establishes |
|---|---|---|
| September 4, 2024 | Public glass-substrate demonstration | An early demonstration, not volume deployment. |
| January 10, 2025 | Pilot line established; mass production targeted from 2027 onward | A planned commercialization path, subject to execution and qualification. |
| September 3–5, 2025 | Glass-core substrate shown with other AI/server packaging technologies | Glass is part of a broader packaging portfolio. |
| November 5, 2025 | MOU with Sumitomo Chemical Group and Dongwoo Fine-Chem to explore a joint venture | Partnership discussions, not a finalized venture or confirmed production schedule. |
As of August 18, 2026, the evidence describes prototype and pilot activity, with mass production planned after 2027—not a glass-core substrate already shipping broadly in commercial AI accelerators. Public demonstrations are an important development milestone, but they do not establish customer qualification, production yields or cost competitiveness (Samsung Electro-Mechanics, CES 2025; Samsung Electro-Mechanics MOU).
Glass is a trade-off, not a cure-all
Greater flatness does not mean a package is automatically more reliable or easier to manufacture. Glass can chip or crack during cutting, handling, drilling and assembly. TGVs must be formed, metallized, filled and inspected at acceptable cost and yield. Defects become especially consequential on large panels, where a crack, void, alignment error or plating flaw can compromise a substantial area.
Rank #4
- 48GB AI graphics accelerator
Nor does glass by itself solve heat removal from high-power accelerators, HBM availability, power delivery, advanced-packaging capacity or the need to qualify a new package with customers. The finished package combines silicon, glass, copper, memory, underfill and other materials; the reliability of that stack depends on how they work together. AI chips still need thermal solutions such as lids, thermal-interface materials, heat sinks or cold plates. New materials and processes may also require new equipment and supply chains.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Other companies are building the glass-packaging ecosystem
This is a broader packaging trend, not a Samsung-only transition. The companies below have different roles, so their announcements should not be read as equivalent products or as proof that glass packaging is already in high-volume use.
| Company | Role in glass-based packaging |
|---|---|
| Samsung Electro-Mechanics | Developing glass-core package substrates and operating a pilot line. |
| Intel | Promoting glass-core substrate technology for advanced packaging. In July 2026, Intel announced a collaboration with Lens Technology on glass materials, laser processing and precision manufacturing. |
| SKC / Absolics | Developing glass substrates for high-performance computing and AI data-center packaging; company performance and commercialization claims should be treated as vendor-reported. |
| AGC | Supplying glass materials and describing TGV substrate capabilities, including structures for chiplet and co-packaged-optics applications. |
| Corning | Offering precision glass carriers for temporary bonding, wafer thinning, fan-out and 2.5D/3D packaging. A carrier is a process material, not a permanent package substrate. |
| Lens Technology | Working with Intel on glass processing and precision-manufacturing capabilities. |
Relevant company materials include Intel’s Lens Technology collaboration, SKC’s glass-substrate overview and SKC’s CES 2025 announcement.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Best Value
- Professional AI & Creator Workstation: AMD Radeon AI PRO R9700 GPU with 32GB GDDR6 is engineered for AI development, professional content creation, and compute-intensive workloads.
- Massive 32GB Memory Capacity: 32GB of GDDR6 memory on a 256-bit bus provides ample bandwidth for large AI models, 8K video editing, and complex 3D rendering.
- Advanced RDNA 4 with AI Accelerators: 64 Compute Units with 3rd Gen Ray Tracing and dedicated 2nd Gen AI Accelerators for groundbreaking AI performance and visual computing.
- Professional Blower Cooling: Efficient single blower design exhausts heat directly out of the chassis, ideal for multi-GPU workstation and server configurations.
- Enterprise-Grade Thermal Solution: Vapor chamber heatsink with industrial Honeywell PTM7950 thermal interface material ensures reliable cooling under sustained professional loads.
What would show that glass is ready for AI production?
A prototype or pilot line is only part of the commercialization story. For a glass package to become a credible production option, chip designers and their manufacturing partners need to validate the complete package under real operating and reliability requirements.
- Is the glass permanent in the final package, or a temporary carrier?
- Does it replace an organic substrate core, a silicon interposer or both?
- What package size, layer count and TGV dimensions and spacing can the process support?
- Are electrical-loss and thermal results measured on a defined structure, and under what conditions?
- What are panel-scale assembly yield, defect rates and cost per package compared with organic substrates and silicon interposers?
- Can the package pass thermal cycling, mechanical shock, reflow and bonding or debonding tests?
- Have customers qualified a design, and is there a production order rather than a demonstration, pilot or MOU?
Those details matter because a material advantage on paper does not guarantee a reliable package at a competitive cost. A substrate can be publicly demonstrated years before a customer qualifies it for a high-volume AI system.
The accurate version of the headline
AI packaging is pushing manufacturers to reconsider the materials around silicon chips. Samsung Electro-Mechanics is betting that glass cores can help meet the demands of larger, more densely connected AI packages. But silicon transistor dies are not going away, and glass-core substrates remain a prospective packaging technology rather than a proven mass-market replacement.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →
