COHR — Coherent, Inc.

The tape is the sentiment here, and as of 2026-10-15 it is ugly: -18.99% in one week, -25.33% over six months, and a ~10% single-session drop in early October driven by rising Treasury yields, higher oil, and open market doubts about whether hyperscaler AI capex is sustainable. Coherent is a high-beta (2.11) proxy for that trade, so it gets sold first when AI infrastructure sentiment cracks. Note the divergence: Street opinion is still constructive - 24 of 31 ratings are Buy/Strong Buy with zero Sells, and consensus targets cluster at $385.89 (all-analyst) to $409.67 (recent). Both reads are partly right: price action is bearish, analyst estimates are still being revised up. One data caveat: the live brief references the stock trading 'in the low-$300s,' while the FMP feed shows $245.09 as of 2026-10-15 - anyone reading this later should anchor to the dated figure, not to the narrative language.

Revenue has risen sequentially for ten straight quarters: $1.053B (Sep-2023) to $1.581B (Sep-2025) to $2.046B (Jun-2026). The important shift is the second derivative - YoY growth ran +17.3%, +17.5%, +20.5%, then +33.7% across the last four quarters, i.e. accelerating, not decelerating. Guidance of $2.2-2.4B for the Sep-2026 quarter implies +39% to +52% YoY, a further step up. Margins confirm it is structural rather than a blip: gross margin expanded from 29.1% (Sep-2023) to 38.5% (Jun-2026), roughly +940bps, and operating margin went from -1.7% to +12.4%. GAAP EPS is noisy (the -$0.83 print in Jun-2025 and -$0.11 in Mar-2025 look like one-off/impairment-adjacent quarters) but the trend in reported net income is clearly up: $240.5M in the June-2026 quarter versus $191.4M and $146.7M in the prior two. The one genuine blemish is cash conversion: operating cash flow of $69.5M in the June-2026 quarter was DOWN 47% YoY despite +34% revenue, and Q3 FY26 OCF was -$93.8M. That is working capital and a capex ramp (capex $562.9M in the June quarter alone, 27.5% of revenue, vs a $90-130M/quarter historical norm), not demand deterioration - but it is real and it is why free cash flow was -$493M, -$383M and -$96M over the last three quarters.

Two very different pictures depending on which lens you use, as of 2026-10-15's $245.09 price. Trailing GAAP: P/E 56.3x on $4.11 of TTM EPS, P/S 6.78x versus a 5-year average of 3.5x, P/B 4.40x. That is what drives the deterministic screen's 'Rich vs. its own history' tag (context score 29.9). Forward: the next-FY EPS estimate is $9.44 on revenue of $10.61B, so the stock trades at roughly 26.0x forward earnings for a business guiding 40%+ revenue growth with expanding margins - not expensive on that basis, and below the 5-year average trailing P/E of 61.1x. The P/S comparison is also somewhat unfair: the 3.5x average was set when gross margin was 700-900bps lower, so a structurally more profitable Coherent should carry a higher sales multiple. Two data flags: the 'dividend_yield_pct' of 5.73% is almost certainly a feed artifact (Coherent does not pay a dividend), and the PEG of 0.03 is meaningless with this EPS base. Net-net: expensive on trailing GAAP, reasonable-to-cheap on forward non-GAAP, and the 5.73% 'yield' should be ignored.

This is a classic estimate-up / price-down divergence. Estimates are moving up (FY27 EPS of $9.44; Q1 FY27 revenue guided to $2.2-2.4B versus $1.58B a year ago; management says 2027 production is 'virtually sold out' and 2028 capacity is being booked), while the stock has de-rated from $440 to $245. Either the market is correctly discounting an AI capex digestion in 2027-2028 that management can't yet see, or the tape is mispricing an accelerating, capacity-constrained order book. The tell is not the price - it's whether FY27 EPS estimates keep climbing into and after the 2026-11-04 print. If estimates rise and the stock doesn't, the multiple is wrong; if estimates stall while the stock is already down 44%, the market was right. Do not underwrite this at 56x trailing earnings; underwrite it at ~26x forward and ask what happens to that multiple if the revenue growth slips from 40% to 15%.

Broken, and not yet washed out. As of 2026-10-15, RSI(14) was 35.64 - weak but above the 30 oversold line - with price -19.2% versus the 20-day SMA, -18.83% versus the 50-day, and -16.76% versus the 200-day, having lost all three in a single week (-18.99%). That is a momentum breakdown, not a range. There is no evidence of a base: recent lows are not being defended, and the 52-week range of $112.32-$440 shows how violently this name can move in both directions. Timing implication: there is no technical confirmation of a bottom here. A constructive setup would require either an RSI flush below 30 with a reversal candle, or a reclaim of the 20-day SMA on above-average volume. Buying at 35.64 RSI with price 19% under the 20-day is catching a falling knife, no matter how good the fundamental story is.

• AI/hyperscaler capex sentiment is the single dominant driver - Coherent's Datacenter & Communications segment is ~79% of revenue, and the early-October selloff was explicitly about doubts over the durability of hyperscaler AI spending.
• Rates and energy: rising Treasury yields and higher oil prices compress high-multiple, high-beta growth names; COHR's 2.11 beta means it amplifies any macro risk-off move.
• InP (indium phosphide) laser fab capacity is the internal bottleneck - management is scaling to double output by late calendar 2027, so revenue is gated by Coherent's own capex execution, not demand.
• Regulatory/trade: proposed FCC moves to block certain Chinese optical transceiver imports would structurally push hyperscalers toward Western suppliers like Coherent - a tailwind - while broader export controls and tariffs are the mirror-image risk.
• Government industrial policy: a $50M CHIPS Act letter of intent supports the 300mm SiC substrate expansion, partially offsetting the capex burden but adding policy/execution dependency.

• Fiscal Q1 2027 earnings on 2026-11-04 - the key question is whether the $2.2-2.4B revenue guide is met or beaten and whether FY27 EPS estimates ($9.44) are revised higher.
• PhotonLink platform and C-band pluggable optical line system ramp, plus delivery against a 2027 book management describes as virtually sold out, with 2028 capacity being booked now.
• InP fab capacity expansion toward doubling output by late calendar 2027 - the unlock for 800G and 1.6T transceiver shipment volumes.
• 300mm SiC substrate customer sampling converting into design wins in AI semiconductor applications, backed by CHIPS Act support.
• Continued institutional coverage initiations/target revisions (Rosenblatt $500 in October; Bernstein and Deutsche Bank initiating Outperform/Buy in late Sept/early Oct) - positive estimate momentum tends to precede price stabilization.

• Free cash flow is deeply negative and getting worse: -$493M (Jun-2026), -$383M (Mar-2026), -$96M (Dec-2025), with Q4 capex of $562.9M equal to 27.5% of revenue. Cash is $1.16B against $3.21B long-term debt - if the AI order book softens while capex is committed, the balance sheet has far less cushion than the 0.325 debt-to-equity ratio suggests.
• Operating cash flow is diverging from reported earnings: OCF fell 47% YoY to $69.5M in the June-2026 quarter even as revenue rose 34%, and Q3 FY26 OCF was negative $93.8M. That is working capital build plus capex, but it means GAAP earnings quality depends on the ramp converting to cash later - if it doesn't, the trajectory breaks.
• Concentration: ~79% of revenue comes from Datacenter & Communications, effectively a handful of hyperscaler customers. A single capex digestion cycle at one or two of them would hit the majority of the business.
• Valuation and multiple risk: at 6.78x sales (vs 3.5x 5-year average) and 56x trailing earnings, with beta 2.11, this stock has limited downside protection if AI infrastructure spending growth merely decelerates rather than stops.
• GAAP vs non-GAAP gap is wide (TTM GAAP EPS $4.11 versus a next-FY non-GAAP estimate of $9.44); much of the forward earnings case rests on adjustments, and a shift in how the market values that gap would re-rate the stock.
• Competitive/pricing pressure in pluggable optics (Lumentum, Fabrinet and Chinese transceiver vendors) plus capacity additions across the industry could turn today's sold-out book into tomorrow's oversupply.

The business is accelerating and the stock is being treated as if it is deteriorating - revenue grew 33.7% YoY with gross margin up ~940bps over three years, FY27 guidance is well above prior Street models, and the forward P/E is ~26x on $9.44 of estimated earnings. But I would not call this a clean buy today: free cash flow is running at -$493M a quarter on a $563M capex program, operating cash flow is falling even as revenue rises, the stock is 19% below its 20-day moving average with RSI at 35.6, and the entire bull case rests on hyperscaler AI capex staying strong through 2028. The honest read is a good, genuinely inflecting business whose valuation is rich on trailing metrics and whose cash conversion has not yet proven the earnings are real - worth owning on a technical stabilization or a further flush, not worth chasing into a broken tape.

I applied a 33x multiple to next-FY (FY2027) estimated EPS of $9.44, producing $311.50 and rounding to $310. The multiple is deliberately between the ~26.0x the stock currently trades at on that same $9.44 estimate and the 40x+ implied by the $385-410 Street targets: 33x credits the accelerating revenue growth (+33.7% YoY, guided to +39-52% next quarter) and the gross-margin expansion from 29.1% to 38.5% over three years, while discounting for the negative free cash flow from the capex ramp, the 2.11 beta, and the genuine risk that hyperscaler AI capex decelerates. It also sits below the company's own 5-year average trailing P/E of 61.1x and the 56.3x current trailing multiple, so it is not a multiple-expansion call - it is a call that forward earnings power, not trailing GAAP, is the right base.