My attitude towards $Advanced Semiconductor Engineering(ASX.US) can be summed up in four words: a good stock, but expensive.

ASE (ASX) is a "second-order beneficiary" in the AI narrative—the more NVIDIA GPUs are sold, the tighter TSMC's CoWoS packaging capacity becomes, and the overflow orders go to second-tier packaging and testing companies like ASE. The first order benefits TSMC, the second order benefits ASE, and the third order only then goes to JCET and Tongfu. So, ASE is essentially in a position of "being lifted by the AI wave." The story holds, but it will never be the most expensive one, nor the cheapest one.

The problem is that it's a bit overdone now. .64, P/E 47x—for a packaging and testing company, this valuation is already pricing it as an "AI leader." The moat of a packaging and testing company is far inferior to TSMC's cutting-edge process technology. This kind of valuation requires 2 years of high growth to digest. One quarterly earnings miss and it will get hammered.

The short-term signals aren't good either. .47 on 5/13 was the recent high, falling to .89 by 5/19, a -7.5% drop in a week. Today it rebounded +3% to close at .64, looking like it's stabilizing, but on 5/18 company executive Jeffrey Chen sold .66M shares—this kind of signal is far more useful than technical analysis. Insiders are selling, why should retail investors rush in.

But I don't want to criticize it too harshly either. The fundamentals are truly good—strong year-on-year revenue growth in April, Q1 earnings beat expectations, and the company even raised its 2026 capital expenditure. AMD announced a $0 billion investment in the Taiwan ecosystem this week, and ASE is also a beneficiary. The story of tight AI backend packaging capacity indeed hasn't broken.

So my stance is:

Don't chase. But don't short either.

If you're already in, take profits on 1/3 to lock in gains, and leave the remaining 2/3 untouched; if you're on the sidelines, wait for a pullback to the 0 range before buying in batches—at that level, the P/E would be around 40x, which is more comfortable than now.

I'm not following those in the group shouting "ASE target 0." A 50x P/E for a packaging and testing company is already the market's extreme faith in AI. Anything beyond that is not a physics problem, it's a religion problem.