Tokyo Electron (ADR)
TOELYQ3 2024(TOELY Q1 FY2025)Estimated14% AI
AI Revenue %
14%
AI Fair Value
$3.6B
AI Revenue (Q)
$525.0M
Total Revenue (Q)
$3.8B
Analysis
Q1 FY2025 (Apr-Jun 2024): Tokyo Electron reported ¥555.07B in quarterly revenue (+41.7% YoY, +1.4% QoQ). AI infrastructure spending was at all-time highs — hyperscalers collectively spending $200B+ annually on AI capex. NVIDIA Blackwell (B100/B200) was entering production at TSMC 4nm, AMD MI350 was in development, and custom AI ASICs from Google, Amazon, Microsoft were in production or qualification. HBM demand was explosive — SK Hynix and Samsung were fully sold out on HBM3E, Micron was ramping. TEL disclosed gross margin of 47.6% (high), partly reflecting favorable product mix toward higher-value AI-related equipment. Management stated AI-related investments were 'increasing significantly due to emergence of AI servers and AI-enabled PCs and smartphones.' TEL's FY2025 target was 30%+ of sales from AI-related equipment (broad definition). For Q1, the broad AI percentage was likely around 28-30%, trending toward the full-year 30%+ target. Applying Ring 1 filter (~50% of broad definition = strictly incremental AI-only demand): ~14-15% of revenue. Ring 1 estimate: HBM etching/bonding (~¥20B, accelerating) + incremental advanced logic for AI accelerators (~¥58B) = ~¥78B of ¥555.07B = ~14.0%.
Analyzed by claude-opus-4-6
AI Products Identified (Ring 1)
Etch systems for HBM3E DRAM layersCoater/developer for 3nm/4nm logic (NVIDIA Blackwell, AMD MI300X)Synapse temporary bonder/debonder for HBM die stackingDeposition systems for CoWoS/advanced packagingCVD systems for GAA transistor nodes (AI chips)Metal etch for HBM TSV interconnectsAerosol cleaning for advanced packaging
AI-Enabled Items (Ring 2 — Not Counted)
These items use AI but are not counted in the AI revenue estimate because they primarily serve non-AI functions.
Equipment for AI-enhanced smartphone SoCsEquipment for AI PC chip manufacturingGeneral data center server chip equipmentMature node tools for AI-adjacent power/networking chips
Confidence Tier
EstimatedNo direct AI revenue disclosure — estimated from product mix