Frontken (0128.KL): The Malaysian Multibagger Cleaning the Insides of TSMC's 2nm Fabs

$0128.KL. One of our multibaggers, with effectively zero coverage on X. Frontken Corporation is a Malaysian company that cleans the chamber components inside TSMC's most advanced fabs. Without them, 2nm chips do not ship. Market cap is 1.9 billion dollars.

That is the entire sales pitch. The rest of this post is about why a Malaysian-listed semicon services company has Cornered Resource economics that very few people outside Hsinchu have noticed.

What Frontken Actually Does

Inside every TSMC fab, deposition chambers accumulate sub-micron contamination with every wafer that passes through. At 2nm nodes, a single 0.3 micron particle is enough to ruin yield on an entire wafer. Frontken takes the contaminated chamber components, removes them from the line, cleans them at the molecular level, and returns them to operation. Each cycle. Every chamber. Every advanced node.

This is the kind of business that does not fit cleanly into any consumer-facing semiconductor story. It is not the chip. It is not the equipment that makes the chip. It is the ongoing service that keeps the equipment that makes the chip actually working. The deeper the node goes (3nm, 2nm, 1.6nm), the more rigorous and frequent the cleaning becomes. 3nm requires 2 to 3 times more cleaning passes than 7nm.

Recurring Service Revenue Tied to Wafer Starts

We have spent meaningful time on the unit economics. This is not a product sale. It is a recurring service revenue stream tied directly to wafer starts. Every chamber cycle produces a cleaning event. Every node transition produces a step-up in cleaning intensity. The revenue scales with TSMC's capacity utilisation, which is structurally near full as long as AI demand keeps the leading-edge order book booked.

Price becomes secondary when yield is at stake. A fab customer running TSMC-grade equipment will not switch a qualified cleaning vendor to save 8 percent if there is any non-zero probability of a contamination event that costs tens of millions of dollars in scrapped wafers. The customer's purchasing decision is risk-managed, not margin-managed. Frontken's net margins reflect that: they run at 26 to 27 percent.

The Moat: Cornered Resource Plus Switching Costs

Think biological symbiosis. Frontken's Ares Green subsidiary sits physically inside TSMC's Hsinchu and Southern Taiwan parks. Not on the perimeter. Inside. Vendor qualification to operate inside a tier-one fab takes one to two years of safety review, contamination protocol audits, and yield-impact validation.

Once embedded, the customer faces a binary choice. Stay with the qualified vendor whose process has been validated against actual wafer yield over thousands of cycles. Or requalify a new vendor over 12 to 24 months while running production. The choice is rarely close. Hamilton Helmer would call this combination Cornered Resource (physical inside-the-fab access) plus Switching Costs (qualified process integration). We covered the broader 7 Powers framework in our explainer on what an economic moat actually is.

"The most underappreciated moats in semicon are not at the foundry. They are at the contractor sitting next to the chamber, holding a wrench."

Two Catalysts in Sight

• Plant 3 Taiwan. An eight-storey facility breaking ground in 2026, designed to double Frontken's Taiwan capacity by 2027. Plus minor Taiwan optimisation works completing in H1 2026. This is direct capacity to capture the next leg of TSMC node migration.
• US fab expansion as the next leg. Frontken is looking at US acquisition targets to crack TSMC Arizona, Intel Foundry, and Samsung. The vendor-qualification moat that protects the Taiwan business in reverse means cracking the US requires either buying an already-qualified incumbent or running the one-to-two-year qualification cycle from scratch. M&A is the faster path. Management is doing the work.

Management on the AI demand backdrop: “The semiconductor business continues to benefit from sustained demand across the artificial intelligence and high-performance computing sectors.” The supply-and-demand setup at the leading edge is exactly where you want a node-leveraged services contractor to be sitting.

Capital Returns Are Modest

The capital returns story is the weakest part of the thesis. Frontken paid roughly 4.0 sen per share in FY2025, a yield of about 1 percent. The payout ratio is about 22 percent. Like most companies listed on Bursa Malaysia, a buyback mandate exists but is barely used.

We would prefer a higher payout. We would prefer some buyback discipline. We accept that for a business reinvesting into Plant 3 Taiwan and exploring US M&A, conserving cash for organic and acquisition capex is the operationally correct call. The capital-return policy is exactly where it should be for a smid-cap mid-cycle compounder, which is also exactly where it gets criticised by income-oriented investors who want more cash today.

The MoatMap Scorecard: Q83 V12 M49

Here is the Frontken MoatMap StockRank:

• Quality: 83/100. Operating margin of 32.4 percent, D/E of 0.03x. The Quality factor loves this profile.
• Value: 12/100. P/E of 45.7x. The market knows.
• Momentum: 49/100. Middling.
• Composite StockRank: 62/100. Mid-pack composite, dragged down by the value factor.

A Q83 V12 M49 profile is the classic glamour-stock shape: extraordinary unit economics, expensive price, average momentum. We covered how to read this kind of factor mix in our guide to factor investing.

The Question Worth Sitting With

Here is the question that keeps coming back. Why pay roughly 51x P/E for Frontken when you can buy TSMC itself at roughly 35x?

The honest answer is that you usually should not, unless you have a specific reason to believe Frontken's growth curve is steeper than TSMC's over the next five years (which is plausible, because services revenue grows with node intensity as well as with wafer count) or that the smaller market cap and lower trading liquidity build in an investor-base discount that closes over time as coverage improves.

The compounder argument is real. The relative-multiple argument is also real. Reasonable people will disagree on which weighs more. The cleanest version of the trade is holding both, sized correctly, rather than choosing.

The Bottom Line

Frontken is a quiet Malaysian semiconductor services multibagger with a structural moat that very few global investors have priced into the multiple yet. Recurring service revenue tied to wafer starts, 26 to 27 percent net margins, Ares Green physically embedded in TSMC parks, Plant 3 Taiwan doubling capacity by 2027, and US M&A in play.

The valuation is rich. The growth runway is real. The moat is genuinely high-quality. The right way to think about a position is the same way you think about any glamour-shaped Quality compounder: long duration, modest sizing, willingness to wear multiple-compression volatility on the way to the eventual rerate.

For investors using Frontken as a single-name idea, our guide to reviewing your portfolio for weak spots is the right framework for sizing Quality-led positions in a portfolio.

For the full breakdown including segment economics, the Ares Green moat in detail, Plant 3 Taiwan capex math, US M&A targets, and the management quality assessment, the Frontken Deep Dive is the place to go.

Disclosure: 0128.KL has been one of the author's multibagger positions historically. This article is for informational purposes only and is not investment advice.