LG Innotek showed a package substrate bigger than eleven reticle-limited dies
The first FC-BGA it has built for AI accelerators runs past 100mm on a side, with mass production targeted for 2027 and glass cores for 2028.

LG Innotek showed its first FC-BGA built for AI accelerators at the KPCA Show in Incheon on September 9th: a substrate more than 100mm on a side, which the company says is about six times the area of the PC substrate it sat next to. Mass production is targeted for 2027.
The chip is not the thing that limits how big an AI package can get. The board underneath it is.
Run the arithmetic on the release and it becomes clearer than the copy. An 85mm square is 7,225mm². A 100mm square is 10,000mm², which is 38% more — and the release separately advertises an "ultra-large-area" sample about 40% bigger than the 85mm part. Those are the same board. Against a maximum single die at the 26×33mm reticle limit, the new substrate is close to twelve times the area, which is the point: this is a carrier for a package of many dies and stacks of memory, not for a chip.
Two other things at the booth matter more than the size.
Chip embedding puts silicon inside the substrate rather than on top of it, which shortens the connection and cuts resistive loss. And glass, which LG Innotek is targeting for mass production in 2028, replaces the core layer to hold the board flat. Warpage is the reason large-area organic substrates are hard. A big ABF board bends as it heats, circuits distort, and yield falls. So the 2028 date is arguably the more important one on this stand, because flatness is what makes area usable.
The quieter number is Cu-Post, which stands copper pillars on the substrate and puts the solder ball on top of them. LG Innotek says this cuts the spacing between solder balls to roughly a fifth of what it was, and that copper conducts heat more than seven times better than the lead it replaces. More I/O, cooler package, same footprint.
Now, is any of this a business yet?
LG Innotek's package solutions division did 935.6bn won of revenue and 99.6bn won of operating profit in the first half of 2026, with margin up from 6.5% to 10.6% and operating profit nearly doubling year on year. The company's stated goal is a trillion won of operating profit from this business by 2031. Annualise the half and you get about 199bn won — so the target asks for roughly five times the current run rate in five years, which is a long way to travel on a substrate that has not been qualified by a named customer.
Because that is the fair objection here. A sample at a trade show is not a design win, and the high end of this market belongs to incumbents. On the Taiwanese side, Unimicron's advanced substrates already sit inside TSMC's 3DFabric alliance (next to CoWoS and SoIC, which is the seat everyone wants). Nobody at LG Innotek claimed an accelerator customer this week.
We would expect one to be named before the end of 2027, or the 2027 production date to move. Watch which of those happens first — and if you follow the packaging supply chain, watch the glass date harder than the area.
