AI-Driven Demand Reshapes Semi Distributor Earnings: WPG and WT Poised for Structural Re-Rating

Core Conclusion

AI infrastructure build-out is generating strong demand for non-GPU components (CPU, networking, memory, PMIC) and AI ASICs, directly benefiting semiconductor distributors WPG Holdings (3702.TW) and WT Microelectronics (3036.TW). Both companies reported 1Q26 earnings materially above expectations, and 2Q guidance signals sustained momentum. Earnings estimates for 2026-2028 are being raised 32% annually for WPG and 12-17% annually for WT, reflecting AI-driven growth that more than offsets weak non-AI end markets. We see 31-47% upside from current levels based on residual income valuations that justify a valuation premium to historical P/E averages.

What the Market May Be Underestimating

The market likely underestimates (1) the scale of AI demand leaking into non-GPU components distributed by these firms—GPU sales are largely direct, but the supporting computer infrastructure (CPU, networking, memory) flows through distributors; (2) the duration of the upcycle, as inference and edge AI demand broaden the base; and (3) the impact of operating leverage at scale, as both firms are seeing rising return on working capital (ROWC) and margin improvement despite product mix dilution toward lower-margin AI hardware.

Evidence Chain

1. WPG 1Q26 earnings beat and 2Q guidance confirm AI-driven acceleration.
1Q26 diluted EPS NT$3.17, +192% YoY and +95% QoQ, 12% above consensus. Revenue NT$316.5bn (+27% YoY) driven by AI infrastructure and supply chain services. Gross margin expanded 0.8ppt YoY to 4.5%; operating margin expanded 1.0ppt to 2.7%. ROWC improved to 15.2% from 11.5% in 4Q25. 2Q26 midpoint guidance implies revenue NT$355bn (+12% QoQ) and EPS NT$3.86 (+17% QoQ), driven by global AI capex and inference/edge AI. Investment implication: The strong beat-and-raise validates that WPG's AI computing exposure (compute segment 49% of 1Q revenue) is sufficient to lift the entire earnings trajectory despite consumer/industrial headwinds.

2. WT Micro's data center revenue share is increasing structurally, driving a multi-year growth runway.

3. Earnings estimates are being revised up materially, signaling confidence in AI-driven growth.
WPG: 2026-2028 EPS each raised 32%, reflecting AI computing growth offsetting non-AI weakness. 2026 revenue growth expected at +34.5% YoY. WT: 2026 EPS raised 12%, 2027 raised 17%, 2028 raised 16%, driven by improved data center outlook and ASIC ramp. Investment implication: The upward revisions are structural, not one-time catch-ups. The magnitude (32% for WPG) suggests prior estimates were too conservative on AI spillover.

Key Disagreements and Risks

• Non-AI demand weakness: If consumer, PC, and automotive segments deteriorate further, WPG's non-compute revenue (currently 51% of mix) could drag. WPG still has high consumer exposure.
• Margin dilution: Both firms face gross margin compression as higher-revenue, lower-margin AI components become a larger share. WT gross margin forecast to drop from 4.04% in 2025 to 3.32% in 2026. Operating margin may also decline if opex grows faster.
• Direct-sales risk: AI GPUs and some ASICs are sold directly by chipmakers, limiting distributor TAM. If direct channel expands to CPU/memory, distributor growth could stall.
• Competition and inventory risk: Rising global semi inventories or macroeconomic slowdown could reverse the cycle. WT bear case (NT$230) assumes data center demand slowdown and share loss.

Valuation or Trading Implications

• New price targets: WT NT$349 (from NT$299), WPG NT$160 (from NT$121), based on residual income models (WT: cost of equity 12.6%, terminal growth 3%; WPG: cost of equity 9.5%, terminal growth 2%).
• Current prices: WT NT$265, WPG NT$108.5 → implied upside of 31.7% and 47.5%, respectively.
• Target multiples (WT 14.2x 2026 EPS, WPG 12.5x 2026 EPS) are above historical averages (WT 8.9x, WPG <10x). This premium is justified by structural AI growth tailwind, sustained earnings revisions, and improving ROWC. Recommend overweight.

Appendix: Key Estimate Changes

*Derived from growth trends in source data.