Long-Term Thesis

Figures converted from Hong Kong dollars at historical FX rates — see data/company.json.fx_rates. Ratios, margins, and multiples are unitless and unchanged.

Long-Term Thesis — 5-to-10-Year View

1. Long-Term Thesis in One Page

The long-term thesis is that Ten Pao is a founder-disciplined Asian power-supply ODM in mid-migration from a 13%-margin consumer-adapter business into a 24%-margin industrial / AI-HPC PSU business, and that over a 5-to-10-year window the protected slice of revenue (smart chargers + lighting + media at ~55% of sales today) compounds faster than the unprotected slice (telecom + new-energy) erodes. The 5-to-10-year case works only if three independent things hold: (i) smart-chargers segment gross margin defends 23–25% as the AI/HPC 3.5–10 kW PSU program qualifies a Tier-1 hyperscaler, (ii) the Mexico/Vietnam/Hungary footprint converts from disclosed-but-unquantified to a measurable 15%+ ex-PRC revenue line that earns a tariff premium, and (iii) Chairman Hung's "walk away from low-margin orders" discipline survives the daughter-led succession that is already staged. This is not a long-duration compounder if any of those three breaks; it is a narrow-moat cyclical ODM with a free option on the Huizhou A-share spin-off. The investor's job over a multi-year horizon is to monitor the protected-slice gross-profit pool, the ex-PRC revenue disclosure, and the related-party leakage trend — not the next quarter's revenue print.

The compounding math is unusually transparent for a HK small-cap: a nine-year average ROE of 22.0% (still 17.5% in FY2025), revenue growth around 10–11% per year over 13 years, capex now digesting after the Huizhou park completion, and a founder who has not sold a share since 2015 sit on top of a 7.9× P/E and a 1.4× book multiple. The asymmetry is in the discount, not the business — but the discount only closes if the protected slice keeps earning its way and the founder family does not extract the upside through the spin-off.

2. The 5-to-10-Year Underwriting Map

Driver #1 — the mix migration into 24%-margin industrial PSU — matters most. It is the single observable variable that translates Ten Pao from a 7.9× cyclical OEM into a low-teens compounder. The other five drivers are necessary conditions; this one is sufficient. If smart-chargers segment GM holds 23–25% for five consecutive years and segment share crosses 45% of revenue, group gross margin clears 20% and the case for a structural multiple rerate writes itself, independent of the Huizhou spin or the succession.

3. Compounding Path

Revenue has compounded at roughly 11% per year over the past 12 years, and ROE has averaged 22% over the last nine. Through the 5-to-10-year window, the path to owner value runs through three mechanical steps: keep growing revenue at the high-single-digit level the industrial mix supports, lift group gross margin from 18% to 20% as the protected slice expands, and let capex digestion convert that into 50%+ FCF conversion as the Huizhou park stops absorbing cash.

The picture matters because it shows the engine working through two completed cycles already. FY2018 was the trough — net income collapsed 65% to $7.0M on a copper-cost shock and the ROE bottomed at 9.5%; revenue still grew. FY2021 was the consumer-electronics peak — net income hit $48.6M on pandemic-era device demand that did not repeat; ROE 28.9%. FY2023 was the destock trough — revenue fell 12% YoY but gross margin expanded 200 bps as the company walked away from low-margin telecom volume; ROE held 19.9%. Through all three, revenue compounded; the variable was margin, and the variable that moved margin was which segment was selling. That is the rhythm to underwrite forward.

The compounding math is asymmetric. Even the base case — which assumes nothing about AI/HPC heroics or the spin-off — models $0.07 of FY2030 EPS at an 11× exit ($0.78 against $0.37, ~16%/yr before dividends, scenario-conditional). The bull case requires evidence currently absent (named AI customer, clean spin economics) but is structurally available if the protected slice keeps earning. The bear case is essentially the FY2018 trough re-run at scale, layered on succession risk. The asymmetry reflects that this is operationally a moat-narrow, finite-runway compounder but valuation-wise a forced-discount situation.

The single most important compounding variable is gross margin — not revenue growth. Adding 200 bps to group GM (18.2% → 20.0%) on FY2025 revenue alone adds roughly $0.013 of EPS at the current effective tax rate. That is more incremental EPS than three years of base-case revenue growth produces without margin expansion. Investors who model this name on "revenue compounds at 10%" miss the lever; the lever is the segment mix.

4. Durability and Moat Tests

A 5-to-10-year thesis only matters if the moat survives the stress tests it has not yet been put through. Two have already been passed — the FY2018 copper-cost shock and the FY2022 smartphone-driven revenue cliff — and the ROE rebuild in both cases is the evidence base for the thesis. Three remain open.

Two tests have been passed already. The FY2018 raw-material shock collapsed net margin to 1.7%; ROE rebuilt to 28.9% by FY2020. The FY2022 smartphone-driven cliff cut revenue 14%; gross margin expanded 200 bps as the company walked away from low-margin volume. Both required no balance-sheet rescue, no equity issuance, no covenant breach. That history is the strongest quantitative evidence the moat exists — a moat that bends, in the language of the Moat tab — and is the reason the thesis can underwrite a 5-to-10-year holding period despite the OEM cyclicality.

5. Management and Capital Allocation Over a Cycle

Founder Hung Kwong Yee built the business in 1979, took it public in 2015, has not sold a share since, and owns 66.78%. The 11-year listed track record — five geographies opened (Hungary 2017, Sichuan/Vietnam 2019, Mexico 2025, with Huizhou Park 2024–25), one major segment created from zero (new energy, 2020), one product platform launched in real time (AI/HPC PSU, H2 2024) — is operationally credible. Story tab gives credibility a 7/10: operational promises are met on roughly the timelines stated, and strategic pivots are telegraphed before being made. That is rare in a HK-listed family business.

What the capital-allocation file actually shows over a full cycle, and what it implies for the long term, is more nuanced:

The clean read on management. On operational execution, this is a B+ founder-operator: every plant said-and-built, every segment built that was promised, mix discipline visible through three cycles, balance sheet net cash, no equity issuance, no covenant trouble, no SFC inquiry, no auditor change. On capital allocation to outside shareholders, this is closer to a C: no buybacks at $0.26, related-party procurement scaling exactly as the spin is staged, daughter on the Nomination Committee that nominates her own appointments, INED tenure 10+ years until the listing-rule cap forces refresh in 2026, and a pre-IPO subsidiary share-award scheme that pre-dates the announced spin by two years. The long-term thesis depends on the second part not deteriorating further; the Golden Ocean Copper cap trajectory ($6.4M → $18.0M in two years) and the Huizhou spin terms are the two specific things that signal whether outside shareholders' compounding is captured or transferred.

6. Failure Modes

The 5-to-10-year thesis has four distinct failure modes. Each is independent — any one triggering would force a re-rate, and any two simultaneously would invalidate the durable-compounder frame entirely.

The Huizhou spin and the mix migration are the two failure modes that matter most. They are not independent — a clean spin that lists Huizhou at A-share multiples is partly conditional on the smart-charger margin holding — but they are observable on different timelines (12–18 months for the spin disclosure, 24–36 months for the segment margin trend), which means a long-term investor can underwrite the position from one signal to the next rather than waiting for both to land simultaneously.

7. What To Watch Over Years, Not Just Quarters

The first of these signals to print is the Huizhou A-share prospectus disclosure of post-IPO economic interest retained by 1979.HK — likely sometime between Q3 2026 and Q2 2027 depending on CSRC processing time. The most durable of the five is the smart-chargers segment GM holding 23–25% with AI/HPC PSU revenue separately disclosed — that is the multi-year evidence that the mix migration is a structural mechanic rather than a cyclical print.

The long-term thesis changes most if the Huizhou A-share prospectus discloses ≥85% economic interest retained by 1979.HK at a pricing range above 25× P/E and the FY2026 H1 interim prints smart-chargers segment GM at or above 25% with a named Tier-1 hyperscaler customer for the 3.5–10 kW AI/HPC PSU program — both conditions together convert this from a narrow-moat cyclical ODM at a forced discount into a structurally compounding industrial-PSU specialist whose 7.9× P/E is the cheapest exposure to AI-server power infrastructure on the HKEX board.

---

Sources: business-claude.md, moat-claude.md, industry-claude.md, numbers-claude.md, forensics-claude.md, people-claude.md, story-claude.md, competition-claude.md, bull-claude.md, bear-claude.md, verdict-claude.md, short-interest-claude.md, technicals-claude.md, research-claude.md. Underlying data from FY2025 annual report (Ten Pao Group, PRNewswire 2026-03-20, HKEX), spin-off announcement (HKEX / BigGo, 14 May 2026), peer financials (Yahoo Finance, 21 May 2026). All figures in US dollars unless noted; converted from HKD at historical FX rates.