Innoscience Poised for 60% Revenue Growth in 2026 as GaN Captures AI Data Center Power Market

Core Conclusion: Innoscience's Gallium Nitride (GaN) technology is critically positioned within the high-value AI data center power upgrade cycle, notably in NVIDIA's prospective 800V HVDC architecture. Management's 2026 revenue growth outlook of ~60% is underpinned by this structural driver alongside automotive and robotics applications, with a clear path to profitability through scale and a higher-margin revenue mix. The primary risk is failure to secure a major role in NVIDIA's future systems. Our 69 HKD price target implies 31% upside, based on 26x 2026e P/S, reflecting high growth ahead of a profitability inflection.

Evidence Chain: AI Data Center Demand and Financial Trajectory

AI data center HVDC power represents a substantial, near-term GaN TAM expansion. The company quantified the GaN opportunity within NVIDIA's planned 800V Rubin Ultra NVL576 system for 2027: a single system requires ~100,000 GaN chips across various voltage levels (8 units of 650V, 116 of 100V, 1,120 of 30V per compute tray), representing ~USD 180k in content value. Innoscience expects to capture leading share against IDM peers (Infineon, Renesas, TI) due to its scale and technological edge, viewing fabless competitors as challenged by yield and reliability risks in production migration. Investment Implication: Success in this design win would materially de-risk long-term revenue and validate the company's IDM model, directly supporting its elevated growth guidance.

Management provides explicit 2026 growth and margin improvement guidance. Revenue is projected to grow approximately 60% in 2026. Gross margin should improve significantly due to economies of scale and a rising revenue contribution from higher-margin data center, automotive, and humanoid robot applications, offsetting consumer electronics weakness. Investment Implication: The guidance sets a high near-term growth benchmark and signals an upcoming inflection in profitability, shifting the narrative from pure top-line expansion to earnings potential.

Long-term capacity and profitability targets demonstrate confidence in structural advantages. Capacity is expected to reach 70k wafers per month (kwpm) by 2028/2029. At this full capacity utilization, the company targets a gross margin of 45-50% and a net margin of 25-30%. Investment Implication: These targets provide a framework for long-term earnings power, suggesting substantial operating leverage as volume scales, which is not yet reflected in current loss-making financials.

Market pricing appears to underestimate the breadth of GaN adoption drivers. Beyond AI data centers, increasing GaN penetration in fast chargers (current 40-50%) and device charging ports (current ~10%) provides a resilient growth buffer against broader consumer electronics softness. Investment Implication: The investment case does not rely on a single vertical; multiple adoption curves support the growth outlook and reduce cycle risk.

Key Risks

• Design-win execution risk: Innoscience may not secure qualification for NVIDIA's 800V HVDC power architecture, or may receive only a minor allocation of the eventual order.
• Competitive intensity: Established IDM competitors or new entrants could engage in more aggressive capacity expansion or pricing, pressuring Innoscience's market share and margin targets.

Valuation & Trade Implication

Our 69 HKD price target is derived from a 26x multiple applied to 2026e sales per share. This multiple is considered appropriate given the company's high-growth phase preceding a clear inflection point from secular drivers like AI data centers. The 31% implied upside supports an Equal-weight rating, balancing the significant growth potential against execution risks and the current pre-profitability profile.