$SSNLF $TSM $AMKR $SKHY 005930 KS 3711 TT Samsung chairman Lee Jae-yong spent Sunday at Samsung's advanced packaging lab in Yokohama. The lab's whole five-year budget is ¥40bn, about $254m, and Tokyo and Yokohama are covering ¥22.5bn of it. ASE spent $991m on packaging equipment in the third quarter alone. That gap tells you what Samsung went to Japan for: materials and process know-how. Capacity is being built elsewhere. Samsung's newsroom says Lee toured the Advanced Package Lab on Nov 1, his first visit since it opened on Sep 8. He went through the lab's work on packaging materials and processes and talked about cooperation with Japanese materials, parts and equipment makers. Samsung names wafer bonding and glass substrates as future research areas. The ¥40bn plan runs from 2024 to 2028 and was first announced in 2023, so the money isn't new. The chairman showing up in person is. Japan's draw is the supplier base around Yokohama. Many of the materials and tools that go into an advanced package come from Japanese companies, and Tokyo is willing to pay more than half the bill to keep that development work at home: ¥20bn from the national government and ¥2.5bn from Yokohama, per Kyodo. Samsung is paying for know-how first. Floor space comes later. Taiwan is where the volume runs today. ASE's packaging equipment spend in a single quarter is close to 4x Samsung's five-year lab budget. TSMC is guiding $60-64bn of capex for 2026, and broker estimates put its CoWoS capacity at roughly 106,000-130,000 wafers a month this year. TrendForce reported on Oct 27 that TSMC is weighing a CoWoS price increase of about 20%. When capacity is tight and concentrated in one place, that place sets the price. The US is building, and the dates all land in 2028-2029. Amkor raised its Arizona plan to about $12bn on Sep 8, the same day Samsung opened the Yokohama lab, with phase 1 due to open in 2028. SK hynix broke ground on a roughly $4bn HBM packaging plant in Indiana in August, first cleanroom in October 2028, backed by up to $458m in federal CHIPS money and up to $712m from the state. TSMC aims to have packaging running in Arizona before 2029. For the stocks, we think Taiwan keeps both the volume and the pricing power in CoWoS-class packaging through 2028, which favors $TSM and ASE. $AMKR's Arizona campus is the biggest US capacity bet, and the revenue from it starts in 2028. Samsung's Yokohama lab is a bet on the next packaging generation, and we don't expect it to move HBM packaging share this year or next. What would change our view: Samsung announcing volume packaging capacity in Japan, or a named Japanese supplier signing a joint development deal with the lab. Neither has happened yet. For the next generation of AI chips, which matters more to you: Japan's materials or Taiwan's capacity?