Fabrinet (FN) Q4 2026 Earnings Call Transcript

Fabrinet reported record Q4 revenue of $1.316B (+45% YoY), exceeding the top end of guidance, and non-GAAP EPS of $4.10 (above the top end of its outlook). Data center revenue rose to $669M (+68% YoY) and reached a $1B annual DCI run rate, while communications infrastructure grew to $413M (+40% YoY). The company guided Q1 FY27 revenue to $1.375B–$1.425B (+43% YoY at the midpoint) with non-GAAP EPS of $4.10–$4.25, alongside strong margin performance (non-GAAP operating margin 10.9%). Management also outlined capacity expansion targeting $12.5B–$14B in potential annual revenue run rate over the coming years.

Analysis

This is less a one-quarter beat than evidence that FN is becoming the toll booth for AI-era optics. The market should focus on operating leverage and capacity optionality: when a contract manufacturer starts adding square footage faster than revenue, the real asset is embedded share of a multi-year customer buildout, not the reported margin print. That favors FN versus slower-moving networking hardware names, while also validating spend at NVDA, AMZN, CSCO, and NOK as their capex is increasingly translated into specialized outsourced production rather than in-house assembly.

Near term, the biggest risk is not demand but conversion: component scarcity, customer timing, and working-capital drag can create visible FCF noise even as revenue accelerates. The inventory build is a double-edged sword; if end-demand stays hot, it is pre-positioned supply, but if any hyperscaler pauses, the market will suddenly reprice this as a cyclical manufacturing name with elevated fixed-cost absorption. The tax provision also matters for headline EPS quality, but it is secondary to whether utilization stays tight into the next 2-3 quarters.

Contrarian view: consensus may be underestimating how quickly FN’s revenue-per-square-foot can scale, which supports a premium multiple, but it may also be overestimating how linear that path is. The key falsifier is a guide cut below ~35% Y/Y growth or any sign that data-center mix growth slows for more than one quarter. If transceiver and DCI ramps are still compounding into late-2027 visibility, the stock likely deserves to keep grinding higher; if not, this becomes a classic high-expectation hardware name vulnerable to multiple compression.

Key Decisions for Investors

• Buy FN on 5-8% post-earnings pullbacks over the next 2-6 weeks; thesis is multi-quarter utilization-driven EPS leverage, with downside if next-quarter revenue guide slips below low-40% growth.
• Initiate a relative-value long FN / short JBL pair for 3-6 months; FN has higher AI-optics exposure and more visible capacity monetization, while JBL is a lower-beta EMS proxy that should not keep up if the market re-rates AI manufacturing winners.
• Add a watch alert on FN if inventory growth exceeds revenue growth for a second straight quarter; that would signal demand pull-forward rather than sustainable ramp and should trigger de-risking.
• Use FN as a basket hedge against AI-networking capex disappointment: if hyperscaler spend remains firm, FN should outperform SMH over the next 6-12 months because the company is monetizing the buildout before peers can.
• If long already, take partial profits only if the stock rerates to a materially richer multiple without a further guide raise; otherwise let winners run until there is evidence of component relief or customer cadence normalization.