Business

Know the Business

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

Ten Pao is a Chinese contract power-supply ODM earning very different gross margins from the same factories: ~24% on smart industrial chargers and AI/HPC server PSUs, ~13% on consumer-electronics adapters, ~8% on EV chargers and energy-storage. At HK$2.91 (US$0.37) it trades at ~7.9× earnings and ~1.4× book against listed peers ranging from 18× to 87× — a discount that compresses three separable issues: HK small-cap with mainland operations, 60% top-5 customer concentration, and a margin mix the market doubts migrates fast enough into industrial/AI-server. The May 2026 announcement that the Huizhou Electronic subsidiary — roughly the smart-charger + new-energy half — is proceeding toward an A-share listing is the capital-structure news that could force a parts-based valuation.

1. How This Business Actually Works

Ten Pao sells the same kind of object — a power-conversion box — into six different markets at six different gross margins. The unit of competition is the design win: an OEM (TTI, Bosch, an unnamed Fortune-500 server builder, an African branded lighting customer, a Chinese EV-charger integrator) hands over a wattage / efficiency / safety spec, multiple Asian ODMs bid, and the winner ships against forecast for 3–7 years. Once a platform is won, revenue is largely volume × commodity cost pass-through; pricing power lives in the qualification asset, not in negotiation room mid-cycle.

Where the real economic differentiation sits in FY2025:

One sentence to remember: the same Dongguan / Huizhou factory earns ~13¢ on every dollar of router adapter, ~24¢ on a smart industrial controller, and ~8¢ on an EV charger — so the only operating metric that genuinely moves group margin is what fraction of revenue is smart chargers + lighting + M&E versus telecom + new energy. Smart chargers stepped from 25.4% margin in FY2023 to 26.7% in FY2024 to 24.1% in FY2025 — that compression, together with new-energy margin falling from 9.3% to 11.3% to 7.7%, is most of why group gross margin dropped 130 bps in FY2025.

The cost stack is commodity-heavy (copper, ferrite, MOSFETs, GaN/SiC for AI-server PSUs) so a quarter of margin moves with the input cycle and a quarter with mix — only the residual reflects management execution. R&D spend lives inside SG&A and is what funded the FY2025 launch of 3,500–10,000W AI/HPC PSUs; those products are how Ten Pao tries to push more revenue into the 24%-margin column.

2. The Playing Field

Ten Pao is the cheapest, smallest, and most under-the-radar of seven listed peers that all make power supplies for some combination of consumer, industrial, EV, and data-center end-markets. The peer set tells you what "good" looks like (Delta Electronics, which has migrated decisively into AI-server PSU) and what the floor is (Phihong, the only one losing money).

What the panel reveals. Among peers with positive ROE, Ten Pao has the second-highest return on equity (17.5%, narrowly ahead of Lite-On at 17.3% and behind Delta at 24.3%) and by far the lowest multiple (7.9× vs 19–88× for the rest). Chicony Power (6412.TW) is the most directly comparable economic substitute — adapter + server PSU, US$1.07B revenue vs Ten Pao's US$0.71B, 16.5% gross margin vs Ten Pao's 18.2%, 14.8% ROE vs 17.5% — and trades at 2.4× the P/E and ~2× the P/B. Delta Electronics is the aspirational peer: it earns 35% gross margin because two-thirds of its sales are industrial / data-center / EV power, not consumer adapters, and the market pays 87× for that mix. The investment question is not whether Ten Pao is cheap on metrics — it visibly is — but whether the mix migrates enough toward Delta and away from Phihong to justify a rerate without the discount-closing event of the Huizhou A-share spin.

3. Is This Business Cyclical?

Moderately, and the cycle hits gross margin first, revenue second. Net margin has swung from 1.7% (FY2018) to 7.5% (FY2016) and back to 6.8% (FY2025) on revenue that grew almost monotonically — meaning the operating picture is far more cyclical than the top line suggests.

Where the cycle actually hits.

• Inputs first. Roughly 45% of cost of sales is copper, ferrite, plastics, PCBs; another 25% is semiconductors. The FY2018 net-margin collapse to 1.7% is the canonical case — copper +27% YoY, MLCC/MOSFET allocation tight, fixed-price OEM contracts. Repeated in milder form in FY2022 (China-shore semis + RMB labour). Mid-cycle gross margin sits 17–19%; trough gross margin is 14% (FY2018).
• End-market mix second. Telecom (router / WiFi) is reasonably steady. Smart chargers / AI-server PSU is hooked to the data-center capex cycle and currently a tailwind. New energy (EV chargers + storage) is the most volatile — Chinese capacity glut pushed segment margin from 11.3% (FY2024) to 7.7% (FY2025), and management is explicitly walking away from lowest-margin orders ("strategic adjustments to optimise customer and product mix … targeting customers and products with lower gross profit margins").
• Working capital third. Receivables $212M and inventory $123M are 47% of revenue combined; that ratio expands in slowdowns and is the channel through which the cycle hits free cash flow. FY2021 saw FCF go negative (–$44M) on capex-plus-WC build into the post-COVID surge.

For a value buyer, the relevant question is whether mid-cycle net margin lives at 6–7% (where it has spent most of the last 6 years) or at 4–5% (FY2017–FY2019 average) once smart-charger margins normalise. The five-year capex digestion at the new Huizhou park (running 5–6% of sales versus a steady-state more like 3%) should support 200–300 bps of operating leverage if revenue grows.

4. The Metrics That Actually Matter

The metrics most investors look at — revenue growth, P/E, dividend yield — are the least useful here. Revenue growth is 3–4% and largely commodity pass-through; P/E is statically cheap because the cycle is real; dividend (4.4% trailing yield) is generous but a result of the spin-off-pending posture, not policy. The single best leading indicator is segment gross margin on Smart Chargers & Controllers — when that line prints above 26%, the next quarter's group net margin rises by 50–100 bps. It dropped from 26.7% to 24.1% in FY2025, which is why the stock dropped 15% on the FY2025 results despite revenue growth.

5. What Is This Business Worth?

The right lens changed in May 2026. Until the May 14 announcement, Ten Pao was best valued as a single Chinese power-supply ODM on EV/EBITDA through the cycle (~5–6× current versus 11–20× peer median = clear discount to peers but rational given HKEX small-cap + customer concentration + cyclicality). After the announcement that the Huizhou Electronic subsidiary will spin off and list on an A-share exchange, a sum-of-the-parts lens is now defensible because the two pieces have materially different multiples available to them: HKEX listed parents trade at 7–10× P/E and Chinese A-share charger / EV-power peers trade at 25–45× P/E.

The mechanical math. At a conservative 20× P/E on Huizhou's ~$28M share of FY2025 net income (scaled to its segment-profit share), the subsidiary alone values at ~$565M — already 1.5× the entire group market cap of $383M. Ten Pao's retained 90%+ stake (the share-award plan covers 9.2% of registered capital) would be worth ~$510M at that multiple. Even at half that valuation, the residual consumer-side listco is effectively free. The variable to track is the spin's actual terms: a successful A-share listing forces a re-mark of Huizhou's economics at A-share multiples; a delayed or failed listing leaves the consolidated HK discount intact.

Three things would reset the valuation conversation:

1. Confirmed A-share filing date and indicative valuation range from the sponsor would compress the discount immediately, irrespective of the eventual listing outcome.
2. Smart-chargers segment margin holding above 25% through FY2026 — this is the operational evidence that the AI/HPC PSU mix is sustainable, not a 2024 one-off.
3. Net-energy segment margin rebuilding to 10%+ — would validate management's "walk away from low-margin orders" strategy and remove the one piece of segment evidence that currently looks worst.

Without those, the stock sets up as a 7.9× P/E cyclical with a free option on the spin-off. With them, the setup is a sum-of-parts where the parts sum to materially more than the whole.

6. What I'd Tell a Young Analyst

Don't anchor to consolidated metrics. Group gross margin (18.2%) is a numerical average of an industrial business earning 24%, a consumer business at 13–25%, and a new-energy business at 7.7%. The investment thesis is entirely about which row you are looking at — when you hear "Ten Pao's margins are compressed," the right reflex is "in which segment, and is that segment the one being spun out?"

Watch the Huizhou Electronic spin-off paper trail. HKEX cleared the proposal under Practice Note 15 on May 14 2026 — first hurdle, not last. The remaining gates are (i) CSRC acceptance of the A-share application, (ii) sponsor pricing range, (iii) lock-up structure for the parent's retained stake. Track those in HKEX filings and Practice Note 15 disclosures. The catalyst that resets the multiple is not completion of the listing but the valuation range disclosed at the application stage — that is when the SOTP arithmetic becomes consensus.

Customer concentration is the easiest way to be wrong. A single customer is 15.1% of revenue and the top five are 59.8%. The FY2025 disclosure mentions Fortune-500 server customers as the new pillar; the FY2024 disclosure was about WiFi 6/7 router customers; the FY2023 about EV-charger OEMs. The customers rotate but the concentration doesn't drop. The risk is not a single customer leaving — it is a qualification slip (yield, delivery, or pricing) on a multi-year platform, which compounds across the top-5 because they all share certain shared suppliers and ODM panel notes. Read the going-concern items and the "key audit matters" in the next annual report for any sign of receivables provisioning rising sharply on a single customer line.

The cheapness is partly real and partly a discount that can close. HKEX-listed small-cap mainland-operations companies trade at a structural discount of 30–40% versus A-share peers; that piece won't close without the spin-off. The cyclical piece (input costs) and the customer-concentration piece are real and won't close without operational evidence. Distinguish which piece of the discount you are getting paid for.

What would change the thesis. (i) The Huizhou spin-off is withdrawn or rejected — kill the SOTP case and value the whole company at 8–10× P/E. (ii) Smart-charger segment margin falls below 22% — the AI/HPC mix shift isn't happening fast enough, value on through-cycle 5% net margin. (iii) Receivables / revenue rises above 35% (currently 30%) — channel stuffing or customer trouble, value on book value at 1.0× P/B. (iv) Chairman Hung Kwong Yee (67) succession — he's the founder and the strategic decisions that brought the company to this point (Mexico, Hungary, AI-PSU, the spin-off) all bear his signature.

---

Figures from data/financials/income.json, segment.json, ratios.json, balance_sheet.json, FY2025 annual report (data/extracted/annual_reports/FY2025_annual_report.md), FY2025 results announcement (PRNewswire, 20 March 2026), spin-off announcement (HKEX / BigGo, 14 May 2026), and peer set from data/competition/peer_valuations.json (Yahoo Finance, as of 21 May 2026).