Coherent's Datacenter Business Powers its Revenue Growth

Coherent's fiscal Q4 2026 datacenter and communications revenue rose to $1.62 billion, up $253 million sequentially and nearly $600 million year over year, driving total revenue to $2.05 billion. The segment represented nearly 79% of quarterly revenue versus about 67% a year earlier, more than offsetting an $80 million decline in industrial revenue. COHR shares have gained 171% over the past year, fiscal 2027 consensus earnings estimates have risen over 60 days, and the company holds a Zacks Rank #2, although its 28.65x forward P/E exceeds the industry's 20.5x.

Analysis

COHR’s investable issue is no longer whether AI-optics demand is growing, but whether its higher-margin transceiver/component mix can outrun the operating leverage and working-capital demands of a rapid ramp. Incremental datacenter revenue should lift gross margin if capacity utilization improves, but the concentration of growth in one end-market raises the cost of any hyperscaler digestion phase; a single quarter of inventory normalization could drive a disproportionate earnings and multiple reset. The article’s segment presentation should be reconciled to the primary filing before trading because its revenue labels are internally imprecise, making headline growth rates insufficient evidence of sustainable profitability.

Relative positioning favors FN as the cleaner picks-and-shovels expression if optical module volumes continue rising: outsourced manufacturing participation can compound without COHR’s exposure to industrial recovery or LITE’s component-supply execution risk. Conversely, COHR has the greatest upside torque if it demonstrates rising gross margin and backlog conversion, but its premium valuation leaves little room for a guide that merely confirms demand rather than raises forward earnings power. Over the next 1-3 months, the key catalyst is forward capacity, lead-time and customer-concentration commentary; over 6-18 months, silicon photonics and higher-speed interconnect transitions could shift value from discrete optical-component vendors toward integrated module manufacturers and foundry-linked suppliers.

Consensus appears to be treating all AI-optics revenue as equally durable. The more relevant distinction is exposure to emergency capacity builds versus normalized replacement demand: a deceleration in unit growth can be masked temporarily by richer speed mix, then surface abruptly through inventory, receivables, or utilization. This makes a broad long basket unattractive after the sector rerating; favor relative-value exposure until independent evidence confirms that orders are shipping through rather than accumulating in the channel.

Key Decisions for Investors

• Initiate a 3-6 month pair trade: long FN / short COHR in equal dollar amounts. FN offers cleaner manufacturing-volume leverage and typically lower technology-transition risk, while COHR carries greater multiple compression risk if margin or industrial recovery disappoints; target 10-15% relative return, exit if COHR raises forward EPS guidance by more than 10% or FN reports customer/inventory deterioration.
• Keep LITE on a post-earnings long watch rather than chase pre-event. Buy only if management demonstrates that supply expansion is not diluting gross margin and guidance implies continued sequential growth after the next quarter; use a 8-10% stop because supply-yield or qualification delays can reverse the AI-optics narrative quickly.
• For existing COHR longs, reduce exposure into strength and retain upside through a 3-6 month call spread rather than unhedged equity. The thesis is falsified by stable-to-rising gross margin, accelerating backlog conversion, and another material upward revision to fiscal-year earnings estimates; absent those, a premium multiple is vulnerable to a 15-25% drawdown on merely in-line guidance.
• Monitor FN receivables, inventory days, and its largest-customer contribution at the next report. A simultaneous rise in inventories and receivables despite strong reported revenue would indicate channel loading and invalidate the preferred FN leg; in that case, close the pair and avoid the optical hardware complex.