Carpenter Technology CRS

Reference table

1. What they actually sell — the end-market split

Carpenter reports by end market. Important caveat: the segment disclosure in this dataset stops at FY2025. The FY26 breakdown is not present, so the mix figures below are one year stale and should be read as structural rather than current.

The structural fact of the whole company: aerospace and defence is not merely the largest market, it is the ONLY one that grew in FY25. Every other end market shrank. That is simultaneously the bull case (Carpenter is levered to the one market that is tight) and the concentration risk (there is no second engine).

Geography (FY25): United States $1.700B (59.1%), Europe $596.2M (20.7%), Asia Pacific $344.4M (12.0%), Mexico $121.5M, Canada $61.2M, other $53.9M. Europe grew fastest at +18.3%. A domestically anchored business with genuine international reach.

Note: FY23 is reported on a different basis entirely — two operating segments (Specialty Alloys Operations $2.214B, Performance Engineered Products $433.7M, less $97M intersegment) rather than end markets. The series is not continuous across that break.

2. The FY26 record — five clean quarters

Carpenter closed FY26 on 2026-06-30 and reported on 2026-07-30. The full-year numbers:

Four years ago this company lost money. The transformation is real and it is documented quarter by quarter.

The quarterly margin sequence is the single most important table in this dive:

Four consecutive beats and four consecutive quarters of margin expansion. Revenue grew 8.6% for the year; operating income grew 34.5%. The earnings growth is overwhelmingly margin, not volume — which is exactly what you want in a supply-tight market, and exactly what has a natural ceiling.

One reconciliation note: the Q2 (Dec-25) diluted EPS in the income statement is $2.09, while the earnings-calendar actual for that date is $2.33. The difference is an adjusted-versus-reported basis. The earnings-calendar series is the one that reconciles to the consensus estimates, so the beat/miss column above uses it; the income-statement figure is used everywhere the dive computes GAAP arithmetic.

Q4 detail worth noting: revenue of $851.0M came in below the $863.3M consensus while EPS of $3.23 came in above the $3.09 consensus. That is a margin beat on a revenue miss — the pattern of a company pricing well into constrained volume. It is a good result, but it is also the shape of a quarter in which the top line is the constraint.

3. Balance sheet and cash flow — pristine, with one asterisk

Balance sheet (FY26, 2026-06-30):

• Cash: $393.3M (up from $315.5M)
• Total debt: $690.7M (all long-term; zero short-term debt)
• Net debt: $297.4M — and against $841.1M of FY26 EBITDA, net debt/EBITDA of 0.35x
• Shareholders' equity: $2.228B (up 18.1%); retained earnings $2.200B
• Current ratio: 3.81 · Working capital $1.461B · Interest coverage 18.6x
• Goodwill $227.3M on $3.838B of total assets — intangibles are just 6.0% of assets, i.e. this is a real-asset business, not an acquisition roll-up

There is no financial risk in this company at present. Leverage of 0.35x turns is about as conservative as a capital-intensive industrial gets.

Cash flow (FY26) — the asterisk:

Operating cash flow grew 37.4%. But capex rose 57.3% — from $154.3M to $242.7M — which held free cash flow growth to 26.6% and, against a $27.0B market cap, leaves a free-cash-flow yield of just 1.34% and a price-to-free-cash-flow ratio of 74.5x. Capex now runs 7.8% of revenue and 1.65x depreciation, meaning Carpenter is investing well ahead of maintenance levels.

This is not a criticism of management — investing into a tight market at 17% return on invested capital is the correct decision, and it is why FY28 earnings estimates look the way they do. It is a caution about valuation. When a stock trades at 41x forward earnings and yields 1.3% on free cash flow, the earnings multiple is doing all the work and there is no cash-return cushion beneath it. The dividend is $40.3M, a 0.15% yield, and rounds to nothing.

4. Data integrity — what I rejected and why

Vendor data corruption is common in this feed and this file has one clear instance plus several inconsistencies worth stating.

REJECTED — the consensus EBITDA series. The estimate block reports ebitdaAvg of $1.483B for FY26 against a reported FY26 EBITDA of $841.1M — an overstatement of 76%. The error is systematic, not a single bad row: FY25 estimated EBITDA is $1.386B against $670.4M actual (+107%), and FY24 is $1.337B against $411.6M actual (+225%). No EBITDA-based forward multiple in this dive uses the estimate series. Where this dive quotes EV/EBITDA it uses reported FY26 EBITDA of $841.1M against my computed enterprise value, and says so.

Cross-check that gives confidence in the rest of the estimate block: consensus FY26 EPS was $10.587 against an actual $10.52 — a 0.6% error. The EPS series is reliable; the EBITDA series is not. That is a useful, specific finding rather than a blanket dismissal.

Checked and found CLEAN — the most recent quarter. inc_q[0] (Q4 FY26, 2026-06-30) passes every test in the standard screen: net income $162.4M against 50.4M diluted shares gives $3.22 versus the reported $3.23; share counts do not duplicate the prior quarter (50.4M vs 50.3M); interest expense of $6.0M is a plausible step down from $8.7M and is consistent with the FY26 total of $37.8M; no sign flips. This quarter is usable.

Checked and found CLEAN — the key-metrics block. enterpriseValueTTM of $27.301B equals market cap $27.004B plus net debt $297.4M, which is arithmetically correct. Net-debt/EBITDA of 0.354x matches my computation of 0.354x. Return on equity of 25.9% and return on invested capital of 17.1% are consistent with the statements. Unlike several other files in this batch, Carpenter's key-metrics block is trustworthy.

Noted inconsistencies (not rejections):

• The ratios_ttm block reports returnOnEquityTTM as null while km_ttm reports 25.87%. Use the latter.
• researchAndDevelopementToRevenueTTM is 0, and the FY26 income statement reports R&D expense of $0 — after reporting $26.1M in FY25, $25.6M in FY24 and $24.4M in FY23. Either R&D was reclassified into cost of sales in FY26 or the field was not populated. Treat the FY26 R&D line as missing, not as evidence that Carpenter stopped spending on research.
• Far-year estimates are extremely thin: FY29 rests on 2 revenue analysts and 1 EPS analyst; FY30 on 2 and 1; FY31 on 1 and 2. Anything past FY28 (5 revenue / 4 EPS analysts) is decoration.
• Segment data stops at FY2025 — there is no FY26 end-market split in this file.

5. Knowledge base — zero claims, and what the adjacent lane says

Carpenter Technology has ZERO tagged claims in the Synthos knowledge base. No tracked source has named this company. Setting kb_claim_count: 0 is an accepted and precedented outcome; this dive is explicitly fundamentals- and arithmetic-driven, and the verdict does not borrow conviction it does not have.

One theme-adjacent claim materially supports the demand backdrop without naming the company:

• All-In, 2026-06-12 (bullish, conviction 58, entity: silver): "The buildout competes for the same critical minerals as space/orbital — nickel superalloys, silver for photovoltaics — so there will be shortages of everything and delays everywhere."

That is a scarcity claim about the exact input class Carpenter converts into finished product, and scarcity is the condition under which a qualified melter earns 22.5% operating margins. It supports the demand and pricing environment. It says nothing about Carpenter's valuation, and we do not stretch it into a conviction rating.

A search-hygiene note. The only other text match on "Carpenter" in the knowledge base is Nassim Taleb, 2020-04-06, on the "carpenter fallacy" — a claim about tail-risk epistemics with no connection whatsoever to this company. This is the same class of error the house has flagged before, where a common surname or noun inflates an apparent claim count. Excluded.

6. Valuation — priced in or room?

At $543.49 (market cap $27.004B, net debt $297.4M, EV $27.301B):

The EBITDA estimate series is rejected (section 4), so forward EV/EBITDA is not computed. FY29 rests on a single EPS analyst.

6a. What today's price assumes (the inversion)

At $543.49 — 41.1x FY27E and 34.1x FY28E — the price embeds the following falsifiable claims:

• The 22.5% FY26 operating margin is a floor, not a peak. Consensus FY28 EPS of $15.92 on FY28 revenue of $3.826B implies a net margin of roughly 21%, versus the 17.0% just delivered — i.e. the price requires margin expansion to continue for two more years beyond a fifth consecutive quarter of it. This is the most fragile assumption in the price.
• Revenue compounds at ≥10% annually through FY29 (consensus: +11.3%, +10.0%, +10.4%), a step up from the +8.6% delivered in FY26 and the +4.3% in FY25.
• Aerospace and defence demand holds through FY28 — with 61.5% of sales in one end market that was the only grower in FY25, there is no diversification cushion if build rates slip.
• The market keeps paying above 30x forward earnings for a specialty metals producer — no de-rating toward the mid-teens multiple this industry has historically carried, despite a company whose operating income was negative as recently as FY22.
• Free cash flow re-accelerates as the capex cycle matures. At a 1.34% free-cash-flow yield, the multiple cannot be defended on cash returns today; it is defensible only if capex normalises and free cash flow converges toward net income.

A near-term data point that cuts against the first assumption: the FY27 Q1 consensus (2026-10-22) is $2.93 against the $3.23 just delivered — a 9.3% sequential decline. Last year the equivalent transition went the other way (FY25 Q4 $2.21 → FY26 Q1 $2.43, up 10%). Either the sell-side is being deliberately conservative, or it has begun modelling the margin peak. That single number is the cheapest available test of the whole thesis.

6b. The return bridge (why the multiple moves)

Expected return ≈ EPS growth + multiple drift + shareholder yield.

• EPS growth: consensus +25.6% (FY27E) and +20.5% (FY28E) — genuinely strong, and the dominant term.
• Multiple drift: our base case assumes COMPRESSION, and this is the whole verdict. We apply 33x to FY28E earnings against the 41.1x the market pays on FY27E today. That is a deliberate de-rating assumption, and the reasons are specific: (i) margin-driven earnings growth is mathematically finite in a way volume-driven growth is not; (ii) a 1.34% free-cash-flow yield gives the multiple no support from cash returns; (iii) the same market rated this company at a fraction of today's multiple when it was posting operating losses five years ago, which establishes the historical range; (iv) the sell-side's own consensus target of $550.60 implies it too sees no expansion left. Note that the roll from 41.1x FY27E to 34.1x FY28E at an unchanged price is mechanical, not re-rating — we count only the drift relative to what the market pays at each date.
• Shareholder yield: dividend $40.3M (0.15%) plus buybacks $179.1M (0.66%) = ~0.8%. Real but immaterial against the valuation question.

The bridge therefore reads: strong earnings growth, largely offset by assumed multiple compression, with a negligible yield underneath — netting to approximately zero over the horizon. That is the arithmetic of a Hold. If we assumed the multiple simply held at 41x, the FY28E fair value would be $654 and this would be a Buy; we decline to assume that, and we say so rather than burying it.

Scenario bands (stated arithmetic, no DCF):

• Bear ~$291 — 22x FY27E EPS of $13.21. Aerospace build rates soften, margin expansion stalls at ~22%, and the multiple reverts toward a normal specialty-metals range. -46% from spot. This is not a remote scenario for a company with an operating loss in living memory.
• Base ~$525 — 33x FY28E EPS of $15.92. Execution continues, margins hold near current levels, and the multiple compresses modestly from today's 41x FY27E. -3% from spot.
• Bull ~$637 — 40x FY28E EPS of $15.92. Margin expansion continues past 24%, the aerospace cycle extends, capex rolls off and free cash flow converges to earnings, and the market maintains today's multiple. +17% from spot. This brackets the $620 street high.

Base sits 3.4% BELOW spot and 4.6% below the $550.60 street consensus. The bear case is 2.7 times larger than the bull case. That ratio is the verdict.

6c. Variant perception (where we differ, what would surprise)

• We have essentially NO variant perception on the fundamentals, and we say so plainly. Our revenue, margin and EPS path is consensus. Per house rule, a name where our view matches consensus defaults toward Hold — and unusually, the sell-side agrees with the conclusion too: a $550.60 consensus target 1.3% above spot is the street's own way of saying "fully valued." When our analysis, the consensus estimates, and the consensus target all converge, manufacturing a differentiated call would be dishonest.
• Where we differ modestly: we weight the cyclicality more heavily than a 41x multiple implies. The consensus curve extrapolates margin expansion; we observe that this company's operating margin has traversed from -1.4% (FY22) to +22.5% (FY26) in four years and treat a 22.5% margin as a cycle-high input rather than a base rate. That is why our bear case is -46% while the street's low target is $425 (-22%). We think the left tail is fatter than the sell-side does.
• Positive surprise that would force a repricing: FY27 Q1 on 2026-10-22 delivering EPS above $3.23 (i.e. above the quarter just reported, versus the $2.93 consensus) with operating margin above 24.3%. That would falsify the margin-peak hypothesis directly and would justify paying today's multiple. Watchable, dated, unambiguous.
• Negative surprise that would force a repricing: the first quarter of operating-margin contraction after five of expansion; or aerospace and defence revenue growth turning negative when the FY26 segment data is disclosed; or a further capex increase that pushes the free-cash-flow yield below 1%.

7. Moat and competitive position

Carpenter's moat is qualification, and it is genuine. High-performance alloys used in rotating jet-engine components must be qualified by the engine manufacturer for each specific application — a multi-year process involving metallurgical testing and regulatory sign-off. Once qualified, a supplier is embedded for the life of the programme, which for a commercial engine can be decades. This is why the company can hold 22.5% operating margins in a business that, at the raw-material level, is a commodity.

Evidence the moat is working in this file: gross margin of 30.6% in a metals business; return on invested capital of 17.1%; the ability to expand operating margin four quarters running while revenue grew only 8.6%; and 138 days of inventory outstanding financed comfortably by a 3.81 current ratio.

Where the moat does not protect: it does not protect against volume. Qualification guarantees Carpenter's share of a programme, not the size of the programme. If aerospace build rates fall, qualified position is worth nothing at the margin — which is exactly what happened in FY21 and FY22, when the company posted operating losses of -$248.6M and -$24.9M. The moat is real and it is a share moat, not a demand moat. That distinction is the entire reason our bear case is -46%.

The peer set in this file (ATI, ITT, Woodward, Mueller Industries, RBC Bearings, MasTec, and others) is a loose industrial grouping rather than a true comparable set. The most directly relevant name is ATI Inc. at $193.12 and a $26.4B market cap — a near-identical size, also aerospace-alloy-driven, also covered in this batch, and trading at 35.6x FY27E versus Carpenter's 41.1x. Carpenter carries the higher multiple and, on the evidence here, deserves it: 22.5% operating margins versus ATI's 13.8%, 0.35x leverage versus 1.89x, and revenue growth of +8.6% versus ATI's +0.6% in its most recent quarter. Carpenter is the better business. It is also the more expensive one.

8. Management, capital allocation and insider activity

Capital allocation in FY26 was straightforward and shareholder-aware: $242.7M of capex (the growth investment), $179.1M of buybacks (up 76% from FY25's $101.9M), $40.3M of dividends (flat for four years), and $7.9M of net debt repayment. Total capital returned to shareholders: $219.4M, or 60.6% of free cash flow. Leverage was reduced. No dilutive equity was issued. This is textbook discipline.

Insider activity — and an important distinction. The July 2026 Form 4 record shows activity from Executive Chairman Tony Thene, CEO Brian Malloy, CFO Timothy Lain and Chief Accounting Officer Elizabeth Socci — but every transaction is either an A-Award (equity grant) or an F-InKind disposal at $576.87, which is shares withheld by the company to cover the tax liability on vesting grants. These are not discretionary open-market sales. Thene received 68,860 shares and had 29,913 withheld; Malloy received 17,216 and had 7,905 withheld; Lain received 18,076 and had 8,300 withheld. In every case the executive's net holding increased.

Read: neutral-to-mildly-positive, and importantly NOT a warning. This is the routine annual vesting cycle. It stands in deliberate contrast to two other names in this batch where officers made large discretionary open-market disposals. The absence of open-market selling at $543 by executives who have watched the stock double is worth noting — though it is weak evidence, and we do not overread it.

9. What would make this a Buy

We want to state the constructive case explicitly, because "Hold" is easily misread as lukewarm and this business is not lukewarm.

Carpenter becomes a Buy on any of:

1. A price in the low $400s — roughly the 200-DMA ($406.75) and ~31x FY27E, at which the bear/bull asymmetry inverts and the margin-peak risk is substantially pre-paid.

2. Operating margin above 25% in FY27 Q1 (2026-10-22) with revenue above $850M, which would establish that the margin trajectory is structural and make 41x defensible on a raised earnings base.

3. Free cash flow above $500M in FY27 as the capex step-up normalises, lifting the free-cash-flow yield toward 2% and giving the multiple a cash-return foundation it currently lacks.

4. FY26 segment disclosure showing aerospace and defence above 65% of sales with continued growth, plus evidence of medical or energy inflecting — a second growing end market would materially widen the moat's demand base.

10. Verdict, kill-criteria and flip conditions

Hold. Carpenter Technology has executed close to flawlessly: four consecutive earnings beats, four consecutive quarters of operating-margin expansion to 24.3%, diluted EPS up 41.8% to $10.52, return on equity of 25.9%, return on invested capital of 17.1%, and net debt of 0.35x EBITDA. The aerospace and defence end market that supplies 61.5% of revenue is tight, the qualification moat is genuine, and management is allocating capital sensibly. None of that is in question.

What is in question is the price. At $543.49 the stock trades at 41.1x FY27E and 34.1x FY28E earnings, 8.7x EV/sales, 32.5x EV/EBITDA and 74.5x free cash flow, after advancing 117.9% in twelve months. Our base fair value of $525 (33x FY28E) sits 3.4% below spot; the street's own $550.60 consensus sits 1.3% above it. The bear case (-46%) is 2.7 times the bull case (+17%). A great company at a fair price with negative skew is a Hold — not a Buy, and not a Sell.

Pre-registered KILL criteria (any one converts Hold → Trim/Avoid):

• The first quarter of operating-margin contraction after the five-quarter expansion streak — the single cleanest signal that the earnings model has changed, given revenue grows only 8.6%.
• Aerospace and defence revenue growth turning negative in the FY26 or FY27 segment disclosure, removing the one growing end market.
• Free cash flow declining year over year in FY27 — capex rising again without a corresponding operating-cash-flow response, pushing the free-cash-flow yield below 1%.
• Net debt/EBITDA above 1.5x for any reason other than a value-accretive acquisition — the balance sheet is currently doing a great deal of work to justify the multiple.
• A decisive loss of the 50-DMA (~$549) followed by the $450 area, which would be the market beginning to price the bear case rather than debating it.
• A shift from tax-withholding disposals to genuine open-market insider selling at these levels.

Pre-registered FLIP TO BUY conditions:

• Any of the four conditions in section 9, most cleanly: FY27 Q1 EPS above $3.23 with operating margin above 24.3% on 2026-10-22, or a price in the low $400s absent fundamental deterioration.

Where CRS fits in the Synthos Framework Portfolio. The industrial / aerospace-supply-chain sleeve, as the quality leg. It is the higher-margin, lower-leverage, better-executing alternative to ATI within the same theme — and, notably, ATI is covered in this same batch with an Avoid verdict. If the sleeve is to be owned at all, Carpenter is the name; the disagreement is about entry price, not about which company. Recommended action: no new capital at $543.49. Existing holders hold with the kill-criteria above armed. Logged as a tracked Synthos call (Hold) as of 2026-08-04 at $543.49.

Single biggest risk: that a 41x forward multiple has capitalised a cycle-peak margin as a structural one. Carpenter's operating margin ran from -1.4% in FY22 to +22.5% in FY26. The market is paying as though the second number is the base rate. History in this very dataset says the first number is also part of the distribution.

Most fragile assumption in the price: that operating margin expands further from 24.3%. Consensus FY28 EPS of $15.92 on $3.826B of revenue requires a net margin of roughly 21% against the 17.0% just delivered. Revenue growth of ~10% cannot produce 20%+ EPS growth without it. Margin is the entire bridge, and margin has already expanded five quarters running.

Provenance & disclosures

• Traceability: ZERO tagged KB claims name Carpenter Technology (kb_claim_count: 0, breadth 0). This dive is explicitly fundamentals- and arithmetic-driven — an accepted, precedented outcome, and not a reason to invent conviction. One theme-adjacent claim is cited and labelled as adjacent: All-In, 2026-06-12 (bullish, conviction 58) on nickel superalloy scarcity. A text match on "carpenter" returning Nassim Taleb's 2020-04-06 "carpenter fallacy" claim was excluded as a false positive.
• Data as-of: fundamentals 2026-06-30 (FY26 Q4 and full year, reported 2026-07-30; fiscal year ends 30 June) · estimates 2026-08-04 · prices 2026-08-04 ($543.49, 50-DMA $549.13, 200-DMA $406.75, RSI 42.4) · KB claims 2026-08-04. Sole market-data source: the Synthos vendor file for this ticker. No figure is drawn from memory or any other source.
• Data REJECTED: the entire consensus EBITDA estimate series. Estimated FY26 EBITDA of $1.483B overstates the reported $841.1M by 76%; the error persists across FY24 (+225%) and FY25 (+107%). No forward EV/EBITDA is computed in this dive. The consensus EPS series, by contrast, was validated (FY26 estimate $10.587 versus $10.52 actual, a 0.6% error) and is used throughout.
• Data gaps explicitly noted: segment (end-market) disclosure stops at FY2025 — there is no FY26 mix; the FY26 income statement reports R&D expense of $0 after $26.1M in FY25, which we treat as a missing field rather than a real change; ratios_ttm.returnOnEquityTTM is null (the key-metrics figure of 25.87% is used); FY23 segment data is reported on an incompatible two-segment basis; far-year estimates are thin (FY29: 2 revenue / 1 EPS analyst; FY31: 1 / 2).
• Data VALIDATED: the most recent quarter (inc_q[0], Q4 FY26) passes the standard corruption screen — EPS reconciles to net income and share count, share counts do not duplicate the prior quarter, interest expense is plausible and ties to the annual total. The key-metrics block is internally consistent (enterprise value = market cap + net debt; net-debt/EBITDA of 0.354x matches my computation).
• Reconciliation note: the Q2 FY26 diluted EPS differs between the income statement ($2.09) and the earnings calendar ($2.33) — a reported-versus-adjusted basis difference. GAAP arithmetic in this dive uses the income statement; beat/miss commentary uses the earnings calendar, which is the series that reconciles to consensus.
• Fair-value caveat: the $291 / $525 / $637 anchors are scenario multiples applied to consensus diluted EPS, not a discounted cash-flow model. The base case explicitly assumes multiple compression from 41.1x FY27E to 33x FY28E, for the reasons set out in section 6b. Assuming instead that the multiple holds would produce a fair value of $654 and a different verdict; we regard that assumption as unsupported and disclose the sensitivity rather than hiding it.
• Not investment advice. Independent research, educational and informational only, never personalised.
• Version: 2026-08-04-full.