Samsung Electronics

This is the first genuinely confusing name in this value chain, so it's worth being precise: Samsung Electronics is not the same company as Samsung Electro-Mechanics, which already has its own profile in stage 2 of this chain. They're both Samsung Group affiliates, but they're pursuing glass substrates for different reasons, on different timelines, out of different divisions — and they're arguably competing with each other as much as cooperating.

Samsung Electro-Mechanics is building glass core substrates to sell to external customers — the same role Absolics and LG Innotek play. Samsung Electronics' glass substrate push instead comes out of its Device Solutions (DS) division, specifically the foundry business that manufactures chips for other companies (Qualcomm, Nvidia, and eventually, it hopes, more of Apple's business). Its interest in glass isn't about selling substrates — it's about using glass packaging to make its own foundry service more competitive against TSMC.

The specific technology Samsung Electronics is chasing is a glass interposer — not a full glass core substrate, but a thin glass layer that replaces the silicon interposer sitting between a chip and the substrate underneath it in advanced 2.5D packaging (the technique that lets a GPU or AI accelerator sit next to HBM memory stacks on one shared base). Its stated roadmap targets 2028 for broader adoption in advanced packaging, following joint R&D with Samsung Electro-Mechanics and Samsung Display — the roadmap that made this Samsung's first confirmed public commitment to the technology.

One detail that separates Samsung Electronics' approach from nearly everyone else profiled in this chain: while Intel and Absolics process glass in large 510mm×515mm panels, Samsung Electronics reportedly works with smaller sheets, 100mm×100mm or less. That's a deliberate trade-off — smaller panels are faster to get into prototype production and qualify with customers, but they're less efficient at real manufacturing scale, since you get fewer chips per sheet. Industry sources read this as Samsung optimizing for speed-to-market over near-term cost efficiency: a bet that being first to show customers a working sample matters more right now than being cheapest to produce.

Unlike every other company in this value chain so far, Samsung Electronics doesn't plan to sell glass substrates or glass processing equipment to anyone. If the glass interposer program succeeds, the revenue shows up indirectly — as more foundry customers choosing Samsung to manufacture and package their AI chips instead of TSMC, because Samsung can offer glass-based packaging that handles heat and warpage better on very large chip designs.

That's a real competitive gap Samsung is trying to close: TSMC's CoWoS packaging platform currently dominates advanced AI chip packaging, and TSMC's capacity constraints there have themselves become a bottleneck limiting how many AI chips Nvidia, AMD, and others can actually ship. If Samsung can offer a credible glass-interposer alternative before TSMC does, it has a real opening to win packaging business it currently doesn't get.

Samsung Electronics sits in stage 4, packaging & assembly, alongside Intel — both are chipmakers using glass packaging to serve their own manufacturing customers, rather than substrate or equipment vendors selling into the chain.

Samsung Electronics is also profiled in The AI Infrastructure Value Chain — stage 03, Memory (HBM).

In May 2026, Korean financial press reported that Apple had opened discussions with both Samsung Electronics and Intel about chip supply diversification, partly because Intel is seen as the furthest along in standardizing glass-substrate packaging (its EMIB approach), fueling speculation that a future Apple chip generation could adopt glass substrates as a standard. Whether or not Apple specifically becomes a Samsung customer for this, the episode illustrates the strategic logic clearly: being early and credible on glass packaging is becoming a way for a foundry to get invited into conversations it might not otherwise be part of.

The bigger structural story is that Samsung is fighting this battle on two fronts inside one conglomerate: Samsung Electro-Mechanics racing to sell glass substrates to Apple, Cisco, and other external customers, and Samsung Electronics racing to offer glass-interposer packaging as a foundry service. They're not the same business, but success in one plausibly reinforces the other — shared component and yield learning inside the Samsung group, even if the two divisions are commercially independent.