AMD and Nvidia face a growing capital burden from substrates without the benefit of foundry margin stacking. Each is committing $1 billion to the substrate ecosystem; AMD has also publicly committed $10 billion to its Taiwan supply chain, more than three times its Q2 net income. That capital is unavailable for R&D and product development, while their gross margins are flat or shrinking. Intel's gross margin is growing as customers prepay its foundry for substrates. x.com 

As custom AI chips become normal, the scarce asset may be the capacity to build them, since every design passes through the same packaging, substrate, PCB material and manufacturing bottlenecks. Meta with MediaTek and Amazon with Alchip show chip design going modular. That favors TSMC for manufacturing, Alchip for ASIC back-end work, Samsung Electro-Mechanics, Ibiden and AT&S for high-end substrates, and Shengyi Technology if AI PCB material stays tight. The bottleneck map widened. x.com  x.com 

IC substrates are Schmid's more relevant growth driver today because demand is already high and rising. Schmid customers Ibiden and Unimicron are expanding capacity with NVIDIA support; higher order intake supports the prospect of more equipment orders. Its balance sheet has improved, and the risk/reward looks attractive at current levels despite volatility. Glass substrates may take another one to two years to matter commercially. This is a personal assessment, not a buy recommendation. x.com  … Show more

Nvidia's co-investment supports Samsung Electro-Mechanics' substrate expansion because it expects capacity to be tight when it needs supply. Samsung Electro-Mechanics says dedicated compute and switch substrate supply to Nvidia starts this year, with double-digit annual volume growth; by 2028 Nvidia orders should take a substantial share of its package-substrate capacity. Global customers fund part of its 6.78 trillion won expansion, but new Sejong output starts in September 2028. x.com 

ABF substrates are among the biggest constraints on AI hardware in 2027, a bottleneck that favors Ibiden and Unimicron. Nvidia or Google can cut HBM per GPU or DRAM per server to ship more, but they cannot work around shortages of ABF substrates and PCB/CCL. x.com 

Samsung Electro-Mechanics is emerging as an AI infrastructure titan as advanced substrates become the bottleneck for AI scaling. Per @jukan05, it will supply FC-BGA substrates for Broadcom's most advanced AI accelerators from the second half, joining NVIDIA, Google, Amazon and Apple. With Ibiden and Shinko at full capacity, it is the only tier-1 player with the capex to scale large-area substrates. AI servers also need 3-4x more MLCCs, where Samsung and Murata own the IP and earn premium margin. … Show more

Ajinomoto, Nittobo, Ibiden and AT&S sit in an ABF substrate supply chain consolidated into near-monopolies, where the AI chip bottleneck keeps moving. Nvidia, Intel and AMD are prepaying for suppliers' factories, with Ibiden receiving over 90 billion yen; companies that size do not prepay unless out of options. Ajinomoto holds over 95% of the substrate film market, and Apple, Nvidia and Google are chasing Nittobo's glass fabric. A capex cycle ending in overcapacity is the risk. x.com  x.com 

Unimicron's profit leverage to rising ABF substrate prices should stay significantly positive and last longer than the market expects, even if customers absorb only part of the increases. As a high-end ABF leader, its high fixed costs and utilization leverage limit any squeeze from costlier T-glass. The T-glass bottleneck and long capacity cycles could extend the benefit for a year or two, and its high Nvidia and ASIC share adds pricing power, though execution on hikes and utilization matters. … Show more

High-end ABF substrates are the next AI hardware bottleneck, giving Samsung Electro-Mechanics a large opportunity, per Meritz Securities. As AI accelerators scale, package areas are growing over 5x and pin counts up to 100x, while Ajinomoto, with over 90% of raw film supply, is maxed out and Ibiden runs at full capacity. Samsung Electro-Mechanics is rapidly expanding FC-BGA capacity to capture unallocated GPU demand, and Nvidia's 2028 Feynman chips may need 3-5x the substrate of 2026 designs. … Show more

AI server PCB and substrate makers are among the safest, lower-risk, lower-reward ways to play Nvidia's racks. PCB value rose 233% in one NVL72 generation, second only to memory, and Morgan Stanley flagged PCB as the top downstream hardware gainer. Unimicron leads AI server PCB with 18-22% share; Zhen Ding gets nearly 70% of revenue from AI; Victory Giant is the most direct Nvidia PCB play; Elite Material's low-loss CCL passed Nvidia certification. TTM, Tripod, Compeq, WUS and Shennan also gain. … Show more