Taiwan: three AI stocks, three very different prices

2026-09-29 · By Lubin Danilo, founder of Lubin Investment

Analyze a stock on Lubin Investment

Three Taiwanese companies never covered here pass all ten criteria of my quality grid: Accton (AI network switches), King Slide (server rails) and Posiflex (point-of-sale terminals). Their prices diverge: Accton trades below my estimate, King Slide far above it, and Posiflex's discount rests on an exceptional cash figure.

Why look at Taiwan beyond TSMC?

TSMC makes the chips, but an artificial intelligence server is also switches that link thousands of machines together, mechanical rails that hold them in the rack, chassis, cooling. A large part of that chain is Taiwanese. I already wrote about TSMC and the Taiwan risk; this time I look at the suppliers nobody mentions.

My screener returns three that earn the maximum score of 10 out of 10. That score measures business quality (margins, growth of cash per share, debt, return on capital), judged separately from price. Price is then judged with P/FCF, the share price divided by the cash the company actually generates each year: a P/FCF of 30 means you pay 30 years of that cash. The details are in my methodology.

Are these three companies worth their price?

The "estimated buy price" comes from my model: it projects cash per share five years out, applies an exit multiple, then brings it back to today by demanding a 15% annual return. A price below it leaves a margin of safety, a price above it leaves none. Three identical scores, then, and three opposite price verdicts.

Accton: the traffic controllers of AI servers

A switch is the traffic controller of a data center: it moves data between thousands of servers at ever higher speeds. Accton builds them as an ODM, meaning for large customers who sell or use them under their own banner. Growth is spectacular: first-quarter 2026 revenue of 70.1 billion Taiwan dollars (TWD), up 64% year on year, then 39.5 billion in July, up 72%, according to Digitimes.

The counterweight shows in the trajectory. Revenue went from 110 to 248 billion TWD between 2024 and 2025, and free cash from 7.0 to 29.8 billion. But the cash margin is 12%, against 19% in 2023: assembling hardware for big customers leaves thin margins, and there are few customers. That is what my resilience indicator captures: 3.5 out of 5, because AI demand is strong but customer concentration limits pricing power.

On price, Accton is valued at 34.7 times its cash, the 23rd percentile of its own history: it has been more expensive about three quarters of the time. The market punishes it for a clear reason, the fear that a pause in spending by the big data center operators would reverse the curve. The fear is legitimate, but it is already largely in the price: my estimate is 2,621 TWD against 1,840 today.

King Slide: the rail with software margins

A server rail is the slide that lets you pull a machine out of its rack. Mundane? Not since AI: an AI server weighs over 70 kg versus under 50 kg before, and liquid cooling shrinks the available space, so rails must carry more in a thinner volume. King Slide holds over 30% of the global market for these rails, which make up 98% of its revenue, and its July 2026 revenue jumped 355% year on year. A North American plant is due to start in September 2026.

The result shows in the accounts: a 56% net margin and a 59% cash margin, levels associated with a software vendor, not a maker of metal parts. The reason is simple: a rail failing in a rack worth hundreds of thousands of dollars costs far more than the rail. The customer pays for reliability, not steel. The risk: Chinese competitors, which pull the resilience score down to 3 out of 5.

The price leaves little room for error. King Slide is valued at 110 times its cash, a level the stock exceeded only 6% of the time in its history (94th percentile). My model assumes cash per share growth of 20% a year, while reality is 51% a year: the assumption is cautious. Even so it gives 5,397 TWD against 12,045 today. Hypergrowth would have to last a very long time to justify that price, and any slowdown would be costly.

What do the revenue and cash trajectories say?

All three roughly double revenue in two years. The difference is what they do with it: King Slide turns more than half of its sales into cash, Accton one eighth, and Posiflex's cash jumped from 1.6 to 4.5 billion in one year, which calls for a check.

Posiflex: the most spectacular discount is also the most suspicious

Posiflex makes point-of-sale terminals: cash registers, touchscreens, ordering kiosks. Activity is rebounding hard: second-quarter 2026 revenue up 46% on the previous quarter, at 5.18 billion TWD, driven by Europe and Asia, with net profit of 467 million against 295 million in the first quarter.

On paper everything screams "cheap": 3.1 times cash, an estimated buy price of 714 TWD for a price of 179, a 6.5% dividend. But look at the gap between 2025 cash (4.48 billion) and net profit (1.99 billion): cash is worth 2.25 times profit. When a company collects far more than it earns, it is very often a one-off release of working capital, meaning customers settling invoices or inventory shrinking. That does not repeat every year.

If Posiflex's normal cash is closer to 2 billion (my inference, not a published figure), the valuation moves from about 3 to about 7 times cash: still low, but no longer the mirage of the first screen. Add a dividend absorbing 87% of profit and a resilience of 0.5 out of 5, because the checkout terminal is a commoditized product. My verdict: a quality company, an opportunity to confirm, not a given.

What three identical scores teach me

The same quality score says nothing about price. Accton is the most interesting at this stage: a good business, valued as if the worst were possible. King Slide may be the best company of the three, and yet the least attractive at the current price. Posiflex teaches caution: a very low P/FCF only means something if the cash producing it repeats.

The shared risks are not trivial: these stocks trade in Taiwan dollars, are not eligible for the French PEA, and depend on the AI spending cycle as well as on strait geopolitics. This is the sorting I wanted to be able to do for any stock, so I built it: the pages for Accton, King Slide and Posiflex detail each criterion. For another country where price diverges from quality, read the Indonesia study.

FAQ

Can Accton, King Slide or Posiflex be held in a French PEA?

No. The PEA is reserved for European companies. These three stocks can only be held in a standard brokerage account, with a broker offering access to the Taiwan exchange, and with currency risk on the Taiwan dollar.

Why is a 10 out of 10 score not enough to buy?

Because the score measures the quality of the business, not its price. King Slide has the maximum score and is valued at 110 times its cash: an excellent company bought too dear can disappoint.

Why should Posiflex's discount be taken with caution?

Its 2025 cash is worth more than twice its net profit, which looks like a one-off release of working capital. On a more normal cash figure the valuation would be closer to 7 times.

Does the geopolitical risk around Taiwan change the analysis?

Yes, it is a risk I cannot quantify and it weighs on every company listed on the island. I treat it as a reason to demand a larger margin of safety, not as a detail.

Related reading

• Indonesia: 4 quality stocks, only one overpriced
• India: 13 quality stocks, wildly different prices
• South Korea: 5 quality stocks the discount still hits

Analyze a stock on Lubin Investment

About the author

Written by Lubin Danilo, founder of Lubin Investment. A self-taught individual investor, I find fundamental analysis fascinating, and it has delivered excellent results. For three years now, my performance has beaten the S&P 500. But analyzing every stock took too much time: sites with incomplete data, calculation methods and criteria never aligned with mine. And spotting the best stocks was just as time-consuming, even with my own well-defined checklist. So I put my software development background to work to build this software, base my investment strategy on its results, and share it with people who share the same passion as me. It judges a company's quality and its price separately, using criteria drawn from the financial literature (Warren Buffett, Michael Mauboussin, Aswath Damodaran).