BRIEF: PCB leader Victory Giant launches $2.2 billion Hong Kong IPO

Victory Giant Technology launched a Hong Kong IPO seeking up to HK$17.49 billion ($2.23 billion) by selling 83.35 million H-shares at up to HK$209.88 each, with trading set to start on April 21. The PCB maker reported 2025 revenue of 19.3 billion yuan, up 79.8% year on year, while net profit nearly quadrupled to 4.31 billion yuan and gross margin improved to 35.2% from 22.7% in 2024. The deal is tied to strong AI demand, with the company citing a No. 1 global market share in AI/HPC PCBs and analysts expecting another 70% revenue growth this year.

Analysis

This is less an IPO story than a capacity-scarcity monetization event in the AI hardware stack. The market is implicitly signaling that PCB is moving from a commoditized mid-cycle component to a strategic bottleneck, and that matters because board supply tends to lag chip demand by 2-4 quarters; that lag can create a temporary supernormal-margin window that attracts aggressive capacity additions, then erodes pricing power later. The biggest second-order winner is likely the upstream equipment and materials ecosystem rather than the board maker itself, because new line buildouts require precision manufacturing tools, specialty laminates, and process-control software.

For NVDA, the implication is mixed but net positive near term: any supplier expansion reduces the risk that board shortages constrain AI server shipments into the next build cycle. The risk is that the IPO becomes a capital allocation signal to the rest of the industry, inviting copycat expansions across Asia; if demand growth normalizes while capacity comes online, the segment could shift from scarcity pricing to competitive bidding by late 2026. That creates a classic setup where the first derivative looks great, but the second derivative of margins turns after the IPO proceeds are deployed.

JPM is a softer beneficiary via fee generation and an improved pipeline for Asia equity capital markets, but the larger takeaway is valuation compression risk for any adjacent AI manufacturing name that has run ahead of the actual order book. The contrarian view is that the market may be over-crediting durability of the current AI infrastructure cycle: if hyperscaler capex grows more slowly than expected, the most levered beneficiaries will see earnings mean-reversion fastest. The setup favors owning the bottleneck enablers while fading the most crowded end-market expressions after the first post-listing pop.

Key catalyst timing is the first 1-2 quarterly updates post-listing, when management commentary on capacity ramp, utilization, and order visibility will reveal whether this is genuine demand pull or simply supply expansion chasing momentum. Any sign that margins plateau before new capacity is fully absorbed would be an early warning that the current rerating is closer to a temporary shortage trade than a multi-year structural revaluation.

Key Decisions for Investors

• Long NVDA into the next 1-3 month pullback only if AI server build guidance remains firm; the trade works as a supply-chain de-risking beneficiary, but trim quickly if board capacity additions start to be discussed broadly across the ecosystem.
• Buy a basket of PCB-capex enablers on weakness over the next 3-6 months; prefer upstream equipment/material names over the issuer itself because they capture the capex wave with less margin-reversion risk.
• Fade the hottest AI hardware adjacencies after the IPO lock-up / first trading-week enthusiasm; use a short basket against long NVDA as a pair if valuations imply permanent scarcity pricing, targeting 15-25% downside if capacity announcements proliferate.
• For JPM, treat this as a tactical ECM fee tailwind rather than a core earnings change; consider a short-duration long into deal-closing windows, but do not underwrite a durable rerating from one transaction.
• Set a 6-12 month alert on utilization and gross margin inflection: if margins stop expanding while capacity is still being added, reduce exposure to the whole AI manufacturing chain because the trade shifts from structural to cyclical.