Novatek's Strong 2Q Guidance Is Transient: Sustainable Earnings and Long-Term AI Gaps Justify Underweight

Core Conclusion

Novatek's 2Q guidance beat is genuine but non-recurring, driven by pull-in demand and KGD price hikes that will normalize by 2027. We retain Underweight because the earnings lift lacks structural durability, and the company's limited AI exposure relative to Mediatek and lost ASIC project constrain re-rating potential. Current 13x 2026e P/E (between historical 11.8x and +1σ 14.8x) leaves no catalyst for premium valuation without secular growth.

Market Mispricing: Transient Beats Mask Structural Erosion

The market is overweighting the 2Q guidance upside (revenue +21% Q/Q at midpoint, gross margin 38-41%) and margin expansion from KGD price hikes, while underestimating three permanent headwinds:

• The pull-in demand is pre-buying, not end-market growth. Consumer-tech visibility into 2H remains limited after early pull-ins from brand OEMs.
• KGD price hikes will not fully offset rising material costs (gold, wafers, substrates) beyond 2026, compressing gross margins from 2027 (modeled at 41.2% by 2028 vs 41.7% in 2026).
• The lost AI ASIC project (explicitly stated) eliminates a potential growth catalyst that peer Mediatek and others retain, creating a secular disadvantage.

Evidence Chain

1Q26 results confirm the transient nature. EPS NT$6.19 (+2% Q/Q, -28% Y/Y) beat consensus by 9%, but revenue NT$23.1bn was exactly in line. The beat came from mix and KGD price hikes (gross margin 39.1% vs consensus 37.7%), not volume-driven demand recovery.

2Q guidance embeds unsustainable pull-in effects. Revenue midpoint implies 21% Q/Q growth, led by TV SoC with KGD price hikes and machine-vision edge AI chips. Yet LDDI will be down Q/Q, and OLED smartphone is flat—no broad-based recovery. Management acknowledged "limited visibility into 2H after early pull-ins."

AI exposure is structurally limited. The company's edge AI opportunity (IP CAM, drones, robotics) addresses a video surveillance TAM growing at only 7.8% CAGR to 2031 (MarketsandMarkets) — too small to offset DDIC competition losses. Meanwhile, the lost AI ASIC project eliminates a direct AI revenue stream.

Valuation is fair but offers no re-rating catalysts. At 13.1x 2026e consensus EPS, the stock sits between the 5-year average 11.8x and +1σ 14.8x. The 4% cash yield provides downside support but cannot justify upper-end multiples without structural growth. The PT of NT$414 (12x 2026e EPS) implies 9% downside from NT$453.

Key Risks and Key Assumptions

Primary risks to Underweight thesis:

• KGD pricing tailwinds persist beyond 2026, extending gross margin expansion above 42%
• DDIC competition abates, protecting market share and pricing power
• Faster 8K/4K TV adoption or foldable smartphone sell-through lifts SoC/OLED DDI revenue
• A new AI ASIC win materializes, closing the AI gap with peers

Downside risks that support Underweight:

• KGD normalization and memory cost increases compress gross margins from 2027 (modeled decline from 41.7% to 41.2%)
• Intensified TDDI/DDI price competition could push gross margin below 40%
• Slower TV or smartphone replacement cycles hurt DDIC volumes
• AI ASIC loss proves permanent, widening valuation discount to AI-exposed peers

Valuation and Trading Implications

The NT$414 target price (12x 2026e EPS) reflects a discount to historical average to account for earnings sustainability risk and AI growth premium gap. With the stock at NT$453, risk-reward is balanced but asymmetrically lower: limited upside catalysts vs. multiple contraction risk from earnings normalization.

Investors should underweight. The 2Q guidance beat provides a tactical exit opportunity, not a reason to add. Wait for evidence of structural demand recovery or AI ASIC project wins before re-evaluating. The 4% dividend yield offers some buffer, but insufficient to compensate for the structural growth disadvantage against Mediatek and other AI-levered semiconductor peers.

Appendix: Key Estimate Changes