The Battle Over Core Layer Materials for Advanced Packaging Substrates: Glass Core vs. Ceramic Core Route Divergence
Nashnova编辑部
AI hardware demand is pushing traditional packaging substrates to their physical limits; the next-generation core layer has split into two competing paths — Intel's glass core and TSMC-plus-Kinsus's ceramic core — both targeting commercialization within three years, with the winner set to reshape the advanced-packaging supply chain.
Why is the current substrate material running out of road?
The mainstream core material today is BT resin — a glass-fiber-reinforced epoxy substrate. As AI chips grow larger and routing layers multiply, warpage and precision are hitting physical walls.
This means → incremental upgrades on the existing material can no longer keep pace; the industry needs a fundamentally different "base plate."
Two replacement paths have emerged: glass core and ceramic core, each backed by a distinct ecosystem.
How far along is Intel's glass-core effort?
Intel has worked on glass-core substrates for over a decade and recently showed the first sample integrated with its own EMIB technology — a bridge that connects chiplets side by side.
Production plan: the Rio Rancho, New Mexico fab will be converted into the world's first glass-core mass-production site; the Chandler, Arizona fab retains pilot-line capability.
The supply chain broadly expects commercialization before 2030.
Core bottleneck: glass is brittle and conducts heat poorly, making processing extremely difficult. In plain terms = the material specs are attractive, but "how not to crack it or overheat it" remains unsolved at scale.
Why did TSMC pick ceramic, and who is building it?
TSMC is co-developing a ceramic-core substrate with Kinsus Interconnect Technology; the internal codename is "EMIB-like" — explicitly positioned as a direct alternative to Intel's glass path.
The formal partnership is currently limited to Kinsus; other major substrate makers such as Unimicron and Nan Ya PCB have limited involvement.
The project is still in early design-verification stage, with a mass-production timeline also beyond three years — roughly parallel to glass core.
Ceramic-core strengths: far better thermal conductivity than glass, more mature processing. Weaknesses: higher manufacturing cost, lower tool throughput.
Where else can ceramic play a role?
Beyond the core layer, ceramic is also gaining traction for the interposer market — the "adapter board" that sits between chips — potentially replacing today's silicon and glass interposers.
Japan's Kyocera and Taiwan's Tong Hsing Electronic are actively pursuing this direction.
This reflects a broader ambition from the ceramic camp: capturing both the substrate core and the interposer layer, not just one.
Could the two materials end up on the same board?
Some supply-chain players are exploring heterogeneous integration — splicing different materials onto a single substrate, each handling its own zone.
In plain terms = glass on the left, ceramic on the right, each playing to its strength — no need to pick one winner.
This means → the final landscape may not be winner-take-all; whoever clears the yield and cost bar first claims the early foothold.
Content is for reference only, not financial advice.