C
Coherent, Inc.
COHR · Technology
Q2 FY2026
Earning ended Dec 31, 2025Report generated Mar 10, 2026
The AI Pickaxe Supplier Priced as If the Gold Rush Never Ends
The market has correctly identified that Coherent makes irreplaceable components for AI infrastructure — but it has priced the stock as if silicon photonics is a distant academic threat rather than an active hyperscaler R&D priority that could commoditize the exact InP transceiver moat driving ninety percent of the bull case.
Flowly Score
Apr 16 Close
-79%$328.00
Fair Value
$70.00
40
Below Averageat $328.00
Business Strength6/10
Genuine compound semiconductor moat — vertical integration into InP fabs is decades-hard to replicate — but ROIC below cost of capital and a compensation structure that rewards the CEO independently of shareholder outcomes keep this from rating higher. The technical depth is real; the organizational and capital discipline is still unproven at scale.
Financial Resilience5/10
The paper losses are accounting fiction masking real operating cash generation, and the Piotroski and Altman scores confirm a business that isn't fragile. But the debt load from the merger still meaningfully constrains strategic flexibility, and the swing from positive to sharply negative FCF in the latest quarter — even as revenue surges — suggests the capex cycle is consuming every dollar the operating engine produces.
Growth & Trajectory8/10
A book-to-bill above four times, data center revenue accelerating ahead of management's own expectations, customer order books extending into 2028, and a six-inch InP wafer transition that quadruples chip output at less than half the cost — this is an extraordinary growth setup, not a normal one. The structural shift from copper to optical interconnects inside data centers is still early, and Coherent is positioned at the exact chokepoint.
Valuation2/10at $252.32
Even the optimistic DCF scenario implies dramatic downside from current prices, and the P/E multiple reflects a market pricing in decades of perfect execution on a business that has never sustained ROIC above its cost of capital. The EV/EBITDA is the more defensible anchor, but the gap between what the business has historically earned and what the current price demands is a chasm, not a gap.
Risk3/10
Silicon photonics at scale is an existential-level threat that could hollow out the InP moat in a single product generation cycle; hyperscaler insourcing programs are live, not hypothetical; Chinese competition is migrating up the performance stack; and the merger debt means a prolonged downcycle hits with no cushion. The near-term visibility is exceptional, but the tail risks are severe and the balance sheet offers no margin for error.
Coherent is a genuinely good business wearing a terrible balance sheet as a costume. The compound semiconductor manufacturing capability — InP fabs, EML lasers, vertical integration from substrate to packaged transceiver — represents the kind of hard-to-replicate physical infrastructure moat that compounds quietly and becomes obvious only after a competitor bleeds capital trying to match it. The AI infrastructure tailwind is structural, not cyclical: every new GPU cluster requires exponentially more optical bandwidth to function, and the addressable market for intra-datacenter optics is still in its early innings. The problem is the price, which requires investors to believe all of this simultaneously: that the AI capex cycle never pauses, that silicon photonics fails to commoditize the transceiver market, that hyperscalers never insource critical photonic components, and that management finally earns returns above the cost of capital on an asset base tripled by acquisition. Any single assumption failing produces dramatic multiple compression. The business is heading toward a genuine inflection — the six-inch indium phosphide transition is a step-change in manufacturing economics, not a marginal improvement, and the 1.6T ramp plus emerging optical circuit switch business gives Coherent multiple growth vectors simultaneously. Order visibility extending to 2028 is unusual in this industry and suggests customers are treating supply security as a strategic imperative rather than a procurement decision. FCF should improve materially as CapEx intensity normalizes post-capacity buildout and the D&A burden from the merger gradually amortizes away from the income statement. The single most dangerous risk is silicon photonics adoption at hyperscaler scale. If the major cloud providers achieve acceptable performance from co-packaged optics architectures built on commodity silicon processes — eliminating the need for discrete InP transceivers — Coherent's core revenue engine faces structural obsolescence in the exact market segment the current valuation depends on. This is not a five-year risk hiding safely in the future; hyperscalers have active programs and massive R&D budgets pointed directly at this problem, and the history of computing suggests that when trillion-dollar companies decide to eliminate a component cost, they usually find a way.