# How Chips Are Using Glass Substrates to Beat Silicon Limits

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 22 Jun 2026, 8:07.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0066.mp3

Episode 66 of The Hardware Podcast. Lucas and Luna explore glass substrates for chip packaging — a material shift that could reshape how processors handle bandwidth, heat, and interconnect density. They discuss why Intel and Samsung are investing in glass core substrates, the manufacturing challenges at scale, and how this technology could enable faster AI accelerators and data-center chips. Specific numbers: glass substrates can reduce signal loss by 30% compared to organic materials and support 50% finer line widths. The hosts explain why glass is suddenly viable after decades of being dismissed as too brittle, and what it means for the next generation of heterogeneous integration. No hype — just the engineering reality. #GlassSubstrates #ChipPackaging #Semiconductor #Intel #Samsung #AdvancedPackaging #HeterogeneousIntegration #AIAccelerators #DataCenter #MaterialScience #Corning #Technology #HardwarePodcast #FexingoBusiness #BusinessPodcast #ElectronicsEngineering #ChipDesign #Manufacturing Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: Luna, I have been digging into something that sounds like a materials-science nostalgia trip but is actually one of the most quietly serious shifts happening in chip packaging right now: glass substrates.
Luna: Glass? Like, what your phone screen is made of?
Lucas: Exactly like that, but we are not talking Gorilla Glass here. We are talking about a thin sheet of specially formulated glass — think Corning or Schott — that replaces the organic resin-based substrate that most chips sit on today. The idea has been kicked around for decades, but it was always too expensive, too brittle, too hard to work with. Now suddenly Intel and Samsung are both publicly investing in glass core substrates, and the timeline is real.
Luna: Before we go deep — and I know you want to — a couple of dollars a month is genuinely what keeps these episodes coming. Buy me a coffee dot com slash fexingo, if you have gotten something useful out of the show.
Lucas: Absolutely. It is a small thing that makes a big difference for us. So, glass substrates — why now?
Luna: Yeah, what changed? I remember hearing about glass packaging years ago and it never made it past the lab.
Lucas: Two things. First, the demand for interconnect density has exploded. When you are trying to pack multiple chiplets into one package — like an AI accelerator with separate compute dies, memory stacks, and I/O tiles — the substrate becomes a bottleneck. Organic materials like epoxy-based laminates start to warp at high temperatures, and they cannot support the ultra-fine line widths you need. Glass is dimensionally stable. It does not expand much with heat, so you can pattern much finer traces.
Luna: How much finer are we talking?
Lucas: Current organic substrates typically handle line widths of around 10 to 12 microns. Glass can go down to 5 or even 3 microns reliably. That means you can route way more signals in the same footprint. And signal loss — glass has a much lower dielectric constant, so high-speed signals degrade less. Early data suggests a 30 percent reduction in signal loss compared to organic materials.
Luna: Thirty percent is huge for something like a data-center GPU where every gigahertz counts.
Lucas: Exactly. And the second reason is that we are hitting a wall with silicon interposers. Silicon interposers are great — they are what enabled the first generations of heterogeneous integration. But they are expensive, they are limited in size by the wafer, and they have their own thermal expansion issues. Glass is cheaper per unit area, can be made in larger panels, and its coefficient of thermal expansion can be tuned to match silicon much better than organic substrates can.
Luna: So it is not just a incremental improvement — it is a structural shift in how you think about the package.
Lucas: Right. The whole package becomes more like a miniaturized circuit board with built-in optical waveguides potentially. One of the really interesting angles is that glass is transparent, so you can align components optically for assembly, and in theory you could even integrate photonic interconnects directly into the substrate.
Luna: But if glass has been known to be better for decades, what was the actual blocker?
Lucas: Manufacturing. Glass is brittle. You have to drill tiny holes — called through-glass vias, or TGVs — to connect the front and back of the substrate. If you try to drill those mechanically, the glass cracks. Laser drilling works, but it was slow and expensive. The breakthrough came from better laser systems and from glass formulations that are less prone to microcracking. Companies like Corning have developed glass with a higher Young's modulus and better fracture toughness specifically for this application.
Luna: So it is a materials science co-optimization problem — the glass chemistry and the laser drilling process had to evolve together.
Lucas: Exactly. And now you have Intel announcing their glass substrate test vehicles for next-generation data-center chips, and Samsung is investing in a dedicated glass substrate pilot line. The timeline that most analysts are pointing to is volume production around 2028 to 2030 for high-end parts. That is not that far away.
Luna: What about cost? Is this going to make chips more expensive?
Lucas: In the short term, yes. The substrate itself might cost more per square millimeter than organic. But when you factor in the performance gains and the ability to integrate more functionality in one package, the total system cost can go down. You might be able to eliminate a separate interposer layer, or reduce the number of PCB layers on the motherboard. Intel has claimed that glass substrates could reduce overall packaging costs by 15 to 20 percent at scale.
Luna: That is the kind of number that gets procurement departments excited.
Lucas: And it is not just about cost. There is a reliability angle too. Organic substrates absorb moisture, which can cause delamination over time. Glass is hermetic — it does not absorb moisture. So for chips that have to survive in harsh environments, like automotive or aerospace, glass packaging is a big win.
Luna: I could see that being a selling point for industrial IoT chips that sit on factory floors.
Lucas: Absolutely. And one more thing I found fascinating: because glass is transparent, you can use optical alignment for pick and place assembly. That could dramatically improve the accuracy of placing chiplets on the substrate, which is a huge yield issue right now. If you misalign a chiplet by even a couple of microns, the whole package might be scrap.
Luna: So glass simplifies the assembly process too. It seems like a classic case of a technology that was ahead of its time, and now the ecosystem has caught up.
Lucas: That is exactly right. The laser drilling tools are mature, the glass suppliers have the right formulations, and the chip designers are desperate for more interconnect bandwidth. It is a perfect storm. I think in five years, glass substrates will be standard for any chip that costs more than a few hundred dollars to manufacture.
Luna: What about the smaller players? Can they adopt glass, or is it only for the Intels and Samsungs of the world?
Lucas: Good question. The glass substrate supply chain is still being built. Corning and Schott are the primary glass suppliers. The process equipment — laser drills, plating tools — is specialized. So early adopters will be the big IDMs and maybe some advanced OSATs like ASE or Amkor. But over time, as the equipment becomes more standard, I expect it to trickle down to the broader ecosystem.
Luna: So where do you see the first commercial products using glass substrates appearing?
Lucas: Data-center AI accelerators, high-bandwidth memory packages, and maybe some high-end networking chips. These are the parts where interconnect density is already maxing out what organic substrates can do. I would not be surprised if we see an announcement from Intel or AMD within the next two years about a specific product using glass packaging.
Luna: I will keep an eye on that. In the meantime, it feels like glass is one of those quiet enablers that most people will never see, but it will make their chips faster and more reliable.
Lucas: That is the best kind of technology — the kind that disappears into the background and just works. Glass substrates are a great example of how the biggest gains in computing are happening not just at the transistor level, but in the packaging that holds everything together.
Luna: Definitely. Thanks for the deep dive, Lucas.
Lucas: Thanks, Luna. And to our listeners — if you enjoyed this, buy me a coffee dot com slash fexingo really does help us keep the show ad-free and independent.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.