[Executive Summary]

1. Current Assessment: “Not a Bubble, but an Early Overheating Phase”

Unlike the dot-com bubble, today’s AI market is supported by real demand — including measurable B2B ROI generation — and massive capital expenditures (CapEx) from big tech companies. While valuation concerns exist in the short term, the market remains structurally bullish due to persistent supply constraints, where infrastructure expansion continues to lag behind demand.

2. Key Driver: Infrastructure Bottlenecks

• Exploding Demand: Big tech (Alphabet, Amazon, Meta, Microsoft, etc.) AI CapEx is projected to reach approximately $650 billion in 2026, up around 60% year-over-year. (Source: Bloomberg)
• Supply Constraints: Expansion of GPUs, data centers, power infrastructure, and cooling solutions is progressing more slowly than demand growth, sustaining strong pricing power across the infrastructure layer.

3. Investment Strategy: Prioritize the “Infrastructure Layer”

Within the AI value chain, the infrastructure segment offers the highest degree of visibility and stability.

• Infrastructure (Preferred): Semiconductors (GPUs/HBM), data centers, power grids, and related areas with highly visible demand.
• Applications (Selective): Focus on Vertical AI and companies with proprietary data moats rather than highly competitive general-purpose services.

4. Key Risks and Risk Management

• Risks: Mid- to long-term oversupply driven by excessive investment, tighter regulations (e.g., the EU AI Act), and shifts in technology standards.
• Response: Use market corrections as strategic entry opportunities while maintaining disciplined risk management through valuation-aware positioning.

5. Conclusion

The core question is not whether to invest in AI, but where to invest. We believe the infrastructure layer — the direct beneficiary of supply constraints — represents the most compelling opportunity, supported by the long-term trend of AI-driven productivity transformation.