TSEM rebounds, FN stalls, same industry, different fate

Some people categorize TSEM and FN as "two semiconductor foundry companies," but honestly—they're not the same type of foundry.

$Tower Semicon(TSEM.US) is Tower Semiconductor, specializing in specialty process wafer foundry: analog chips, power management, RF, CMOS image sensors, silicon photonics. It's the "specialist foundry" besides TSMC—TSMC does the most advanced 2nm/3nm, TSEM does less advanced but highly profitable mature processes, like automotive-grade ICs, power ICs, sensors. It processes silicon wafers.

$Fabrinet(FN.US) is Fabrinet, specializing in optical component EMS foundry: doing precision component assembly for optical communication companies—optical modules, lasers, fiber connectors, thermal management components. It doesn't process chips, it does assembly. Its biggest customers are NVIDIA's subsidiaries, Cisco, Coherent, these optical module manufacturers. To put it simply, TSEM is "making chips on silicon wafers," FN is "assembling chips with lenses, fiber optics, and circuit boards into products."

Why does the market put them together? Because the AI optical communication trend has pushed them both into the spotlight—TSEM makes silicon photonics chips (the $1.3B big order on 5/14 ignited the concept), FN does optical module assembly (Morgan Stanley called it a data center bandwidth arms race). The more AI computing power is stacked, the greater the demand for optical interconnects, both are "behind-the-scenes supply chain beneficiaries."

But the performance on 5/20 alone shows their "different fates":

TSEM rose 5.25% to close at $264.48. The weekly trajectory was quite dramatic—on 5/14, Q1 earnings with net profit +62% YoY + Q2 revenue guidance hitting a new high pushed it directly to $280; on 5/18, it suddenly dropped -9.4% in a single day to $248; on 5/19-5/20, it rebounded to $264. Market sentiment is being pulled back and forth between "AI silicon photonics" and "high valuation."

FN fell 2.29% to close at $662.21, down -4.8% for the week. After the Q3 earnings on 5/4, it surged over +16.2% to $746, but then declined for 4 consecutive trading days. The problem lies in the cautious Q4 Outlook, plus that "Don't Thank Nvidia" analysis on 5/19—FN's growth doesn't rely on direct NVDA orders, but on data center assembly foundry, with low unit prices and thin margins.

My judgment is:

TSEM is better than FN, for three reasons.

First, stronger institutional consensus: TSEM has 4 Strong Buy + 1 Hold + target price +18.66%; FN has 4 Strong Buy + 3 Buy + 2 Hold + target price +13.12%. Both are bullish, but TSEM's consensus is more concentrated.

Second, more solid earnings delivery: TSEM Q1 net profit +62% + Q2 guidance hitting new high + $1.3B silicon photonics order; FN Q3 Beat but Q4 Outlook cautious. One is accelerating, the other is decelerating.

Third, valuation and growth are better matched: TSEM's PB 10x looks expensive, but it can hold up in the silicon photonics track (automotive / analog / silicon photonics, three lines with +20%+ growth); FN's PE 56x is on the high side in the "EMS foundry" track—the gross margin ceiling for optical assembly is just 12-15%, it doesn't deserve such an expensive PE.

If you want to bet on the AI optical communication behind-the-scenes chain, my own play is to wait for TSEM at the current price of $264 to pull back to $250 and buy in batches, target $300+; I would watch FN first, wait until the Q4 guidance is revised, at least don't enter until it's below $700.

I could be wrong—FN's position as the "optical module EMS leader" is hard to replace, and there's still room in the long term. But in terms of relative strength in the short term of 1-3 months, TSEM is more favorable than FN.