Business

Know the Business — Tripod Technology Corporation (3044.TW)

Tripod is a one-product Taiwanese factory operator: it sells printed circuit boards, and 99.6% of revenue comes from that one line. The whole investment case is therefore a question about what kind of boards — and the answer is changing fast, because in the last two years the mix has tilted from PC/handset commodity multilayer toward 16–32-layer AI-server, high-speed networking and 800G optical boards. That mix shift, not volume growth, is what drove gross margin from 19.3% in FY2023 to 25.9% in FY2025 and ROE from 14.1% to 19.6%. The market is correctly pricing the move (trailing P/E around 25x) but probably underestimating two things: how clean Tripod's balance sheet is for a mid-cycle factory operator (net cash equal to roughly 9% of market cap, zero meaningful long-term debt), and how disciplined capex still is even at peak (FY2025 capex NT$5.96B vs depreciation NT$4.29B — barely above run-rate).

How This Business Actually Works

Tripod converts three inputs — copper-clad laminate (CCL), labour, and large stocks of depreciating equipment — into multilayer printed circuit boards, then sells them by the square foot to contract manufacturers (Foxconn, Quanta, Wistron) and OEMs in PC, server, networking, automotive, and consumer end markets. It is a manufacturing toll: revenue is per-area-times-complexity, the cost stack is roughly half raw materials, and the lever every PCB CFO pulls is mix — moving square feet from cheap 4-layer commodity boards to expensive 16–32-layer server / HDI / 800G optical boards.

Two mechanical points an investor must internalise. First, depreciation does not flex with revenue, so utilisation is the dominant short-cycle margin driver — Tripod's fixed-asset turnover went from 2.65x in FY2023 to 3.57x in FY2025, and that single metric explains most of the 660 basis points of gross margin expansion. Second, the OEM holds end-pricing power and the CCL oligopoly upstream holds material pricing power, so the only sustainable margin lever Tripod controls is which boards it chooses to make. The 0.42% of sales spent on R&D is not a moat — it is a process-engineering line item. The competitive position rests on scale, qualifications, plant geography, and discipline, not on patents.

The Playing Field

Tripod sits at #8 on the Prismark global Top-20 ranking — fourth-largest in Taiwan after Zhen Ding, Unimicron, and Compeq. The right way to read the peer table is not to ask who is largest, but who has chosen which slice of the PCB stack. Zhen Ding lives off Apple flex/HDI and is barely profitable right now (3.7% net margin) because handset/wearable mix has rolled over; Unimicron is an ABF-substrate house masquerading as a PCB peer (market pays 31x forward earnings because the substrate cycle is a different business); TTM is the US-listed aerospace-and-defence beneficiary; Gold Circuit is the closest mix match — a Taiwan AI-server PCB pure-play earning 16.0% net margin and trading at 65x trailing earnings.

Against that backdrop Tripod looks like a mid-mix value name: lower margin than Gold Circuit, higher margin than Zhen Ding or Compeq, balance sheet better than any of them, and trading at the lowest trailing P/E in the group. Whether that is a bargain or a fair discount depends on how durable the mix shift turns out to be.

Note: TTMI figures converted to NT$ at 2026-05-21 spot rate of NT$31.57 per USD; gross margin and ROE blanks for the Taiwan peers are Yahoo data gaps, not zeros. Source: company filings, Yahoo Finance, Prismark Top-20 (BofA 2026 conference deck).

The bubble chart is the cleanest way to see the dispersion. Tripod sits at the lowest P/E in the group at the second-highest net margin — only Gold Circuit converts revenue to profit at a higher rate, and the market pays nearly three times the multiple for it. The reason is that Gold Circuit is a near-pure AI-server PCB business, while Tripod's mix still includes handset, automotive, and PC boards that are not riding the same up-cycle. The bull case is that Tripod re-rates toward Gold Circuit as mix continues to shift. The bear case is that handset/PC drag prevents that re-rating and Tripod stays a "mid-tier value" name.

Is This Business Cyclical?

Yes — moderately to deeply. PCB is the second derivative of consumer and enterprise electronics demand, and Tripod's own revenue printed -10.5% in FY2023 (the cleanest recent downturn) before recovering +11.8% in FY2024 and +11.5% in FY2025. Gross margin compressed only about 110 basis points peak-to-trough (FY2022 17.9% to FY2023 19.3% — note FY2023 actually rebounded slightly from FY2022 thanks to early mix shift) which is far better than the 200–400 basis points typical for the industry. The reason is that Tripod was already moving away from PC/handset mix when the cycle broke and Bien Hoa Vietnam came online in 2023 as a tariff hedge rather than capacity-add at the wrong moment.

The investor takeaway: Tripod is cyclical but better-cushioned than its peers, and the cushion comes from three structural choices — (a) staying out of substrates (no Unimicron-style capex hole), (b) early Vietnam diversification, and (c) maintaining net cash through the cycle. Trough earnings power matters more than peak earnings power for valuation; Tripod's trough was FY2023 EPS of NT$11.53, which the stock now trades at 42x — peak earnings power is much higher.

The Metrics That Actually Matter

Five metrics carry most of the explanatory weight. P/E and dividend yield are not on the list because, like every PCB name, Tripod's earnings move 30–60% peak-to-trough and headline ratios are nearly useless mid-cycle.

Capex discipline deserves its own line. Industry-average capex/depreciation for the Taiwan PCB names ran 1.5–2.0x during the 2021 build-out; Tripod has stayed at 1.4x even in FY2025. That is the difference between earning a mix-shift bonus and giving it back to the cycle.

What Is This Business Worth?

The right lens for Tripod is normalised earnings power times a through-cycle multiple, not SOTP. There is genuinely one economic engine here — 99.6% of revenue is PCB fabrication. There are no listed subsidiaries to discount, no holding-company structure, no regulated/non-regulated mix, and no separable asset value worth pulling out. The only segment disclosure ("PCBs 99.63%, Other 0.37%") confirms that.

The interesting question is therefore not how to value Tripod but what to normalise to. The choices are: trough FY2023 earnings (NT$11.53 EPS, ROE 14.1%) priced at 42x, peak FY2025 earnings (NT$19.45 EPS, ROE 19.6%) priced at 25x, or a normalised mid-cycle EPS the analyst computes themselves. The market is implicitly pricing about 80% credibility on FY2025 being closer to the new normal than FY2023 was.

The cleanest cross-cycle anchor is EV/EBITDA: 13.6x — the lowest in the peer set, against a band of 15x (Zhen Ding) to 49x (Unimicron). Combined with the second-highest net-margin profile in the group, that gap is the re-rating space if mix-shift gains hold. The downside anchor is P/B: at 4.7x against a trough P/B around 2x, a meaningful chunk of the mid-cycle re-rating is already in the price. A buyer here is paying for the mix shift to be permanent.

What I'd Tell a Young Analyst

Don't model revenue — model gross margin and ASP per kft². PCB top-line growth is the volume-cycle echo and explains less than mix does. If you can explain the gross-margin trajectory you have the stock; if you only have a revenue forecast you do not. Track FY2026 actual volume against the 76,000 kft² company target — if revenue grows but volume hits target, the mix shift is intact; if revenue grows on volume only, ASP is rolling over and a margin warning is coming.

Watch capex / depreciation, not capex absolute. Headlines about "NT$10 billion 2026 capex for Vietnam expansion" sound alarming but are inside a depreciation base that may also be rising — what matters is the ratio. Above 1.6x for two consecutive years is the cycle-killer; Tripod has not been there in over a decade.

Use the peer dispersion as the valuation guardrail. Gold Circuit at 65x trailing earnings and Compeq at 47x define the band Tripod can re-rate into; Zhen Ding at 69x and Unimicron at 208x are not relevant because they are different businesses (flex/HDI and ABF substrate respectively). The relevant comp band is roughly 25–65x, with Tripod entering at the bottom; mix-shift durability tells you where it ends up.

The thesis can change on three signals: (a) Tripod's quarterly gross margin reverses by 200+ basis points without a one-off explanation; (b) a single new customer rises above 15% of revenue (creating an AI-server cliff); or (c) capex/depreciation steps to 1.6x or higher and stays there. Until then, this is a clean way to own the AI-server PCB mix shift without paying for it twice.

One final note on what this business is not. Tripod is not a moat business in the durable-competitive-advantage sense. The 0.42% R&D ratio, the lack of patents that matter, and the absence of any platform or network effect should make that clear. The competitive position is qualifications, scale, geographic diversification, and capital discipline — repeatable but not unique. Pay accordingly.