The Take — Great business, wrong price. Even after a 45% drawdown, you're paying for the bull case.
The operating story is real and getting better. FY26 revenue +23% to $7.12B, Q4 +34% Y/Y to $2.05B, datacenter & comms +40% to $5.27B. Clean operating margin (SBC expensed, acquisition amortization and one-offs stripped) went 8.2% → 15.1% → 17.7% over three years, and Q4 ran 19.2%. Q1 FY27 guide ($2.2–2.4B, 39.5–41.5% non-GAAP GM, opex $400–420M) implies another step up. Street FY27/FY28 revenue has been raised to ~$10.6B / ~$14.8B. This is no longer the broken II-VI integration — it's the #2 merchant optics supplier riding the copper-to-optical shift in AI clusters.
But growth is consuming every dollar it earns. FY26 operating cash flow was just $80M against $1.10B of capex → FCF of roughly −$1.0B. Inventory went $1.44B → $2.58B; payables more than doubled. Street has FY27 capex ~$1.75B and FCF ~−$700M. The company funded it with a $2.0B equity raise (share count 155M → 197M basic / 202M diluted after the preferred conversion). Per-share value is being built with shareholder capital, and on a goodwill-inclusive, R&D-capitalized base the business earns an ROIC of only ~8.5% today — below its 11.5% cost of capital.
A generous base case still lands at $164. Revenue +45% / +30% / +20% to $16.1B by FY29 (more than double today), R&D-adjusted EBIT margin rising to 25% (≈21.5% clean, ≈24% non-GAAP — above the 21.8% Q4 print), 12% terminal ROIC. That's worth $164.07. Give it the full Street revenue path (~$19.3B FY29) and 29% margins and you get $268 — only 10% above today. You need the blue-sky case ($20.8B FY29, 33% margin — Innolight-class economics on a company with a 26% industrial drag) to justify the June high near $440.
Reverse DCF: at the base growth path, $243 requires a ~35% R&D-adjusted margin forever; at a 25% margin it requires 1.4× base growth every year for a decade. That's the AI-optics capex cycle never mean-reverting — in a component business with Chinese competitors (Innolight, Eoptolink) and hyperscaler buyers who dual-source by design.
What changed since the Aug-11 build ($73 base at $325): that model used a 15% margin on ~$6.6B TTM revenue with no R&D capitalization — too punitive given Q4's print and the raised Street path. This rebuild more than doubles base value to $164. The stock also fell 25% (from $325 to $243). The gap closed from both sides, but it's still 48%.
Call: AVOID new money above ~$190. Accumulate <$165 (base). Aggressive starter ≤$185 (prob-weighted $183). Holders: this is a momentum/AI-capex position, not a value hold. Size it as a cyclical trade (≤2% of capital), and trim rallies toward $280–300. Rates are the near-term tape driver: the 10/8 selloff (−9.6%) and today's −7.7% are yield/AI-capex sentiment moves, not company news. Q1 FY27 print Nov 4 after close. If expressing a view, do it defined-risk: a post-print bear call spread above ~$300 (the 10/8 breakdown level) captures rich IV. Don't short it naked — beta ~2, and one hyperscaler 1.6T award can gap it 15–20%. If you want to own it, sell cash-secured puts at $170–180 strikes: you get paid for the volatility to wait for a base-value entry.
Risk Profile — HIGH
HIGH
AI-capex cycle / hyperscaler concentration. ~74% of revenue is datacenter & comms, driven by a handful of hyperscaler and GPU-platform buyers. Long AI cluster build-outs; short any capex digestion pause (optics demand is a derivative of GPU shipments and falls harder).
HIGH
Pricing / competition. Chinese transceiver makers (Innolight, Eoptolink) lead share at 800G/1.6T. Co-packaged optics and LPO could shift value to switch/GPU vendors. ASP declines of 15–25% a year are the industry norm; margin expansion depends on InP laser in-sourcing and mix.
HIGH
Reinvestment burn. FCF ≈ −$1.0B in FY26 and ~−$0.7B expected in FY27. Inventory has nearly doubled to $2.58B, which carries obsolescence risk if 800G→1.6T transitions faster than expected. More equity raises are possible.
MED
Rates / multiple. Long-duration cash flows: 61% of base value is terminal. WACC ±100bp moves value ~21%. The October drawdowns have tracked the 10-year yield (above 5%).
MED
Trade / China. Significant manufacturing in China and Malaysia, plus China customer revenue. Tariffs and export controls cut both ways.
LOW
Balance sheet. $2.0B cash and investments against $3.2B debt (net debt ~$1.2B, under 1× EBITDA). The preferred has been eliminated. No solvency risk.
What the "inventory" really is: $2.6B of transceiver components and finished modules tied to one technology generation, plus an InP fab build-out. You are long the AI optical attach rate and short ASP erosion. The tornado's widest bars are the growth path (how long the AI build-out lasts) and the discount rate. Margin is a close third. This is a duration bet on the AI capex cycle.
Scenario values per share vs price
Bear: FY27 +30%, then a FY29 digestion year (−5%), 16% margin. Base: $16.1B FY29, 25% R&D-adj margin. Bull: Street path ($19.3B FY29), 29%. Blue-sky: $20.8B FY29, 33%. Prob-weighted 20/50/20/10 = $183.35.
Green = above price $243; yellow = within ±5%; orange/red = below. The price sits on the 1.0× path only at ~35% margin, or at the 1.4× path (FY29 ~$21B, above Street) at 25%.
Sensitivity: value/share by WACC (high-growth → stable) × terminal growth
Invested capital (book, incl. $4.4B goodwill) + research asset
$12,790 + $1,751 = $14,541
R&D-adjusted after-tax ROIC (19% tax)
8.5%
*FY22–24 one-offs (restructuring, impairments, integration) were not broken out in the XBRL pull, so clean EBIT for those years is GAAP + acquired amortization only. That understates FY23–24 (the Coherent Inc. integration years). Normalization note: the 5-yr average (12.6%) gives normalized EBIT of $900, which values COHR at well under $100. I did not use it as the base because this is a secular ramp, not a cyclical round-trip. It is the floor if AI optics mean-reverts.