WinWay Technology: CPO Testing Bottlenecks Create a Multi-Year Socket Opportunity

Core Conclusion

WinWay Technology (6515.TW) is positioned as the primary beneficiary of the CPO testing equipment cycle, but the near-term revenue driver remains traditional test sockets. The CPO transition introduces new optical alignment and module-level test bottlenecks—sub-micron fiber alignment and repeated FAU connections at scale—that WinWay addresses across insertion points 3/3.5/4. However, the market may overestimate CPO’s near-term revenue contribution while underestimating the multi-year testing complexity that supports socket value growth. Stock trades at 111.5x 2026E P/E, pricing in aggressive adoption that leaves limited upside absent execution beats.

Evidence Chain

1. CPO roadmap requires fundamentally new test infrastructure beyond traditional probe cards. CPO scales from pluggable optics at 800G to 2.5D/3D CPO targeting 12.8T+ by ~2028–2030, per WinWay's roadmap. Fiber arrays expand from 8x/16x to 64x/160x links for future XPU optical I/O. Single-mode fiber with a 9 μm core demands sub-micron alignment tolerance; FAU (fiber array unit) tolerance stack-up reaches 3.8 μm, making optical loss a yield-limiting factor. As fiber counts rise to 64–128, repeated alignment and FAU connection steps slow module-level testing dramatically. Investment implication: CPO test speed and accuracy become new bottlenecks that cannot be solved by existing probe-card architectures, necessitating specialized socket and handling solutions—a structural volume driver for WinWay's test socket business as CPO adoption ramps from 2027–2028.

2. WinWay’s competitive focus should remain on test sockets, not system-level equipment. WinWay offers solutions across CPO test insertions: EIC probe card (Insertion 1), optical/electrical test sockets and double-sided probing (Insertions 3/3.5), and HyperSocket (Insertion 4). HyperSocket combines spring-probe with elastomer contact to reduce contact resistance, improve current capacity, and lower Joule heat. However, double-sided probing systems remain experimental and do not directly compete with MPI/TEL on die-to-die probers. Investment implication: WinWay’s core edge lies in its socket design for optical/electrical hybrid testing—a higher-value, higher-margin niche than probe cards. The CPO/CPC testing interface across insertions 3–4 represents a natural extension of WinWay's existing socket franchise, not a pivot into unfamiliar equipment territory.

3. TSMC’s advanced packaging roadmap (larger interposers, glass substrates) supports CPO scaling—but adoption timelines are long. TSMC disclosed 14-reticle interposers (20x HBM) by 2028 and >14-reticle (24x HBM) by 2029; SoW enables interposers beyond 40 reticles, integrating up to 64 HBM stacks. Glass substrates offer better CTE matching with silicon, lower electrical loss, and stability—critical for large CPO packages. Investment implication: TSMC’s packaging scaling directly increases the addressable market for WinWay’s test sockets (larger packages = more test points per unit). But these are 2028+ catalysts; near-term (2026–2028) copper-optical coexistence via CPC (Co-Packaged Copper) solutions fills the void, providing a transitional revenue source while CPO yield ramps.

Key Divergence & Risks

Market perception vs. reality: Consensus likely lumps WinWay with general “AI test equipment” beneficiaries, but CPO-specific socket content is still in early qualification. Revenue contribution from CPO remains negligible in 2026–2027; 2028 EPS estimates of NT$348 imply steep growth that depends on CPO adoption hitting inflection.

Downside risks:

• Slower-than-expected CPO yield improvement delays socket orders beyond 2028.
• HyperSocket or MEMS probe card adoption falls short of forecasts.
• Edge AI demand fails to materialize, reducing overall test demand.
• Alternative optical coupling technologies (e.g., direct fiber attach) could reduce testing complexity.

Upside risks:

• Faster CPO adoption by hyperscalers drives 2028 revenues ahead of estimates.
• CPC or HyperSocket learns migrate into traditional socket upgrades, raising ASP.

Valuation / Trade Implication

At NT$10,540, the stock trades at 111.5x/58.2x 2026E/2027E P/E and 92.2x 2026E EV/EBITDA. The price target of NT$12,000 (14% upside) is derived from a residual income model (9.14% cost of equity, 16.4% medium-term growth, 4.5% terminal growth). Current valuation embeds aggressive assumptions—2028E P/E of 30.3x is more palatable but still prices in a smooth CPO adoption curve. The 1Q26 earnings call on May 26 is a key catalyst; modest upside surprise is possible given HyperSocket momentum. For long-term investors, the risk/reward is balanced around CPO timing. Near-term entry point is unattractive unless 2026 EPS guidance is raised materially. Wait for pullback to 70x forward P/E or evidence of CPO socket design wins before adding positions.