Over the past 7 days, Lasertec has lost 12% of its market cap. The sell-off is noise. The ledger points to a structural bid that most traders are ignoring.
Goldman Sachs recently issued a bullish note on three Japanese semiconductor equipment firms—Lasertec, Tokyo Electron, and Disco—citing Intel's decision to raise its 2026 capital expenditure by approximately $3 billion. The core thesis: Intel's aggressive push toward 18A and 14A nodes, combined with its EMIB-T advanced packaging technology, will drive a multi-year procurement cycle for high-end wafer fabrication and packaging tools.
But the market misread the signal. The immediate reaction was muted, with analysts questioning whether a $3B increment justifies the premium valuations. I audited the order book proxy data. The real story is not the marginal CapEx increase—it's the structural shift in Intel's supply chain. Intel is betting its IDM 2.0 future on three Japanese suppliers because they own the critical bottlenecks: Lasertec controls ~85% of EUV mask inspection, Tokyo Electron dominates coater/developer (>50%), and Disco holds 50-80% of the precision dicing/grinding market for chiplet packaging.
The Contrarian Angle: Intel's execution risk is real, but the equipment orders are already locked in. Retail traders focus on the possibility of Intel delaying 18A. Smart money looks at the lead times. High-NA EUV tools from ASML require 12-18 months for delivery and ramp. Lasertec's inspection tools must be integrated with those scanners. Tokyo Electron's etch and deposition chambers are spec'd into Intel's fab plans months before the first wafer is run. The $3B CapEx increase likely reflects signed purchase orders, not optional guidance. Canceling now would incur penalties that Intel, with its $80B+ in committed CHIPS Act subsidies and tight cash flow, cannot afford.
The systemic root-cause analysis reveals a second layer: advanced packaging. Disco's equipment is essential for EMIB-T bridges that connect Intel's compute chiplets with HBM stacks. The AI inference market is shifting toward heterogeneous integration. Even if Intel's 18A node fails to match TSMC's N3 in performance, the packaging demand for AI accelerators remains inelastic. Disco benefits from both Intel's and TSMC's packaging roadmaps—it is a double dip.
Where the market gets it wrong: The sell-off in Japanese equipment stocks after the initial Goldman report was driven by fear that the US would force Intel to prioritize American suppliers like Applied Materials or Lam Research. This is a mispricing of political risk. The CHIPS Act grants require domestic manufacturing, but not domestic equipment sourcing. Japan is a trusted ally; its firms are already embedded in Intel's R&D roadmaps. The US Commerce Department has no incentive to disrupt a functioning supply chain that produces the $2 trillion AI market.
Takeaway: The probabilistic setup favors long positions in Lasertec and Disco over the next 12-18 months. Entry levels near ¥68,000 for Lasertec and ¥55,000 for Disco offer a 1:3 risk-reward if Intel maintains its current timeline. The key signal to monitor is Intel's Q1 2025 earnings call—any mention of 18A yield progress above 70% will trigger a re-rating.
"Skepticism is the only viable alpha." The market will price in Intel's failure before it happens. Until then, buy the equipment, not the hype.
