Nittobo's T-Glass supply crunch deepens despite 3.6-fold capacity expansion plan for AI chips

The T-Glass crunch is not a capacity story—it is a structural mismatch. AI compute demand has yanked a niche material track (historically single-digit CAGR) into an exponential curve. Nittobo's 3.6x expansion will not materialize until FY2029, locking in a tight-equilibrium regime for advanced packaging substrates through 2028. Technical cascade: Low-CTE glass fiber cloth is the load-bearing element in ABF substrates and silicon interposers, directly governing warpage control in 2.5D/3D packaging. Any delivery slip at Nittobo hits OSATs in Taiwan, China first, then ripples into NVIDIA and AMD shipment cadence. The deeper risk: fewer than two qualified global suppliers exist, and qualification cycles run 18 to 24 months. There is no Plan B in the near term. Competitive dynamics: NEG and Ajinomoto will almost certainly accelerate vertical-integration negotiations to lock in priority allocation. Chinese glass-fiber players have scale in commodity electronic cloth but remain a generation behind in low-CTE specialty drawing—effective substitution is 12+ months away at best. 12-24 month outlook: Expect a replay of the 2021 silicon-wafer shortage—40%+ annual price escalation, 3-5 year take-or-pay contracts from hyperscalers, and mid-tier packagers forced onto degraded alternatives. Nittobo's ¥20B investment is, in substance, a capital moat that converts a supplier into a price-setter, extending Japan's material-monopoly playbook (photoresist, silicon wafers) into the packaging era.