The framing from Merck's electronics chief is worth taking seriously precisely because it comes from a materials supplier, not a foundry. His argument: as transistor scaling slows, the differentiation moves upstream and sideways — into the materials chemistry, advanced packaging, metrology, and the software that stitches these steps into a working process. Calling AI-era progress "Moore's Law on steroids" is less a claim about density and more about the shape of competition changing.
The strategic implication for the global industry is a shift from vertical process leadership to horizontal integration capability. For two decades, advantage concentrated at the leading-edge foundry that could print the smallest features. If Merck is right, the next decade rewards whoever can co-optimize dozens of interdependent variables across the stack. That favors deep, long-term supplier partnerships over transactional purchasing, and it explains why packaging players, EDA vendors, and materials houses are suddenly strategic rather than commoditized. The risk is fragmentation: no single actor controls the full chain, so execution now depends on collaboration quality — a softer, harder-to-defend moat than a process node.
For Japan, this thesis is unusually favorable. The country's semiconductor position eroded at the fabrication layer but stayed dominant in materials, chemicals, and precision equipment — exactly the layers Merck says are becoming decisive. If competitive advantage migrates toward integrating materials, packaging, and metrology, Japanese suppliers sit closer to the value than they have in years. Rapidus and the broader domestic push gain a credible supporting cast rather than having to rebuild everything from scratch.
The caveat for Japanese firms is organizational, not technical. "Collaboration as the next Moore's Law" demands cross-company co-development, shared data, and fast iteration — historically not a strength of siloed Japanese supply relationships. The opportunity is real, but capturing it means treating integration and software orchestration as first-class capabilities, not afterthoughts to hardware excellence.
For SIers and enterprise dev teams, the read-through is indirect but concrete: the AI compute buildout that drives this materials demand also tightens hardware lead times and pricing. Procurement and capacity planning for AI infrastructure should assume a supply chain where advanced packaging, not just chip availability, becomes the bottleneck.
