Graham Corporation is a classic good-pond/small-fish story. With ~$210M in annual revenue against US Navy submarine programs running in the tens of billions, it occupies sole-source or near-sole-source positions on specific fluid, thermal, and turbomachinery sub-systems where switching costs are measured in years of re-qualification, not dollars. The three-platform structure — Graham Manufacturing (vacuum/heat transfer, Batavia NY), Barber-Nichols (turbomachinery, Arvada CO), and FlackTek (advanced mixing, Louisville CO) — gives product diversification within a tightly bounded niche. The moat is real: classified/ITAR-controlled designs, multi-decade program relationships, and proprietary engineering. The structural weakness is the mirror image of that strength: ~85% of the $515.6M backlog is defense, concentrated in a single service branch's nuclear shipbuilding account. At $104/share (~$1.16B market cap), the stock trades at a substantial premium that prices in not just the backlog ramp but a margin inflection the company has never demonstrated at scale.

Graham Corporation (GHM) — Deep Dive

Graham Corporation is a precision-engineered, sole-source supplier of three types of critical sub-systems:

1. Vacuum and heat-transfer equipment (Graham Manufacturing, Batavia NY) — steam surface condensers, ejectors, and heat exchangers used on US Navy aircraft carriers and in industrial/energy processes.
2. Turbomachinery (Barber-Nichols, Arvada CO) — pumps, turbines, compressors, and expanders for US Navy nuclear submarines, commercial space launch vehicles, and energy applications.
3. Advanced mixing (FlackTek, Louisville CO, acquired January 2026) — bladeless dual asymmetric centrifugal (DAC) mixing technology used in defense, space, and industrial composites.

The business is not a commodity pump manufacturer. It makes classified or ITAR-controlled, program-specific components where the engineering qualification takes years to establish — making switching costly and slow. Graham does not compete for spot contracts. It wins long-cycle sole-source program positions and then supplies replacement and aftermarket parts for the program's lifetime.

Management has articulated a three-part strategic agenda:

1. Margin expansion to low-to-mid-teen EBITDA by FY2027 — from the current ~10.8% nine-month FY2026 adjusted EBITDA run rate, implying 200-500 basis points of additional expansion through operating leverage as the $515.6M backlog converts.
2. 8-10% annual organic revenue growth — funded by the Virginia-class and Columbia-class submarine program ramps, growing commercial space cryogenics orders, and FlackTek's defense/industrial mixing revenues.
3. Platform diversification — FlackTek is the first deliberate acquisition to reduce US Navy concentration, adding a third manufacturing platform with exposure to non-Navy defense and industrial markets.

The T. Rowe Price $50M equity placement (April 2026) is stated to fund debt repayment and future organic/inorganic growth, suggesting additional M&A optionality is in management's thinking.

Revenue (EDGAR)

Operating Income and Net Income (EDGAR)

FY2025 GAAP operating margin: 7.2% ($15.2M / $209.9M). Net income FY2025 is the highest since FY2015 — a ten-year gap between meaningful profitability episodes.

Cash Flow and Balance Sheet (EDGAR)

Note: OCF fell from $28.1M (FY2024) to $24.3M (FY2025) despite operating income more than doubling ($6.9M to $15.2M) — working capital consumption is absorbing operating leverage gains, typical of a defense contractor ramping production. The $35M FlackTek acquisition closed January 2026 against $21.6M of year-end cash; the T. Rowe Price $50M equity injection (April 2026, 8-K confirmed) substantially relieves near-term liquidity pressure.

Share Count (EDGAR)

Cumulative dilution: ~8.8% over the decade from FY2014 to December 2025, plus approximately 5.5% from the T. Rowe Price placement. The placement was at $83.36, approximately 20% below the then-prevailing market price, creating immediate book dilution for existing shareholders.

Current multiples (at $104.15, ~$1.16B market cap):

• Trailing P/E: ~95x (FY2025 net income $12.23M)
• GAAP operating margin FY2025: 7.2%
• EV/EBITDA: ~44-45x on FY2026 guidance midpoint ($26M adjusted EBITDA)
• P/S: ~5.5x trailing revenue

Peer context:

• Ducommun (DCO, ~$900M market cap): 17.5% EBITDA margin, trades at a fraction of GHM's EV/EBITDA
• Curtiss-Wright (CW, $10.5B): 16.6% adjusted operating margin, lower multiple despite superior margins
• HEICO (HEI, $26B): 20%+ EBITDA, 30-40x EV/EBITDA — the only peer trading at GHM-like multiples, and only after decades of proven mid-teen-plus margins

Fair-value range (transparent method): Using a bull-scenario FY2027 outcome: 13% adjusted EBITDA on $275M revenue = ~$35.75M EBITDA. At 25x EV/EBITDA (a premium niche-defense multiple, not HEICO-level), EV = $894M. Adding estimated net cash of ~$30-40M post-T. Rowe Price proceeds and dividing by ~11.6M post-placement shares yields approximately $80-90/share. At 35x EV/EBITDA (HEICO-like premium, requiring sustained execution): ~$115-120/share. The current $104.15 sits between these bounds, requiring roughly 30-35x EV/EBITDA on not-yet-achieved FY2027 targets. The consensus analyst 12-month target was $81.67 (late May 2026, 5 analysts, MarketBeat/DailyPolitical) — already 21% below current price.

Fair-value estimate: $80-120 range; central case $85-95 on FY2027 execution at 25-30x; downside to $65-75 on margin disappointment.

• T. Rowe Price: Purchased 599,808 shares at $83.36 for $50M in April 2026 (8-K confirmed), approximately 5% of common equity. Largest single institutional event in recent history. Positioned as a long-duration holder.
• Analyst coverage: 5-6 firms, predominantly buy-rated. Northland upgraded to Outperform in February 2026; Zacks upgraded to Strong Buy in April 2026. Thin coverage signals underfollowedness, not over-hype.
• Dilution trend: Shares grew approximately 8.8% over the decade FY2014-December 2025, primarily from equity compensation and the FY2022 equity raise. The T. Rowe Price placement adds another ~5.5%.
• No evidence of insider selling concentration, paid promotion, or activist pressure. Short interest at 3.68% of float (403,906 shares, May 15, 2026) with 2.1 days to cover — moderate, not crowded. Month-over-month increase of +29.6% suggests some institutional skepticism ahead of June 8 earnings.

The thesis: Graham holds contracts that cannot be replaced. The Virginia-class follow-on ($136.5M, through February 2034) and the Columbia-class MK19 contract (eight-year stream entering a legislated production ramp) together represent more than $150M in contracted, sole-source revenue with a single customer that has no substitute. Operating income went from -$11.3M (FY2022) to +$15.2M (FY2025) on $87M of incremental revenue — a 30% incremental operating margin (EDGAR-confirmed arithmetic). If the $515.6M backlog converts with normal labor utilization and the Columbia ramp proceeds as authorized, the FY2027 low-to-mid-teen EBITDA margin target is achievable through pure operating leverage, not heroic cost-cutting. T. Rowe Price's willingness to pay $83.36 — below the then-market and well below current levels — for 5% of a $1B company provides real institutional validation. Commercial space cryogenics represents a second growth vector with zero legacy competition at Graham's engineering depth; the $22M in FY2025 orders from six customers is early but real.

What must go right: FY2027 EBITDA margin reaches 13%+; Columbia-class production actually ramps to 1 boat/year without Electric Boat labor/supply-chain slippage; backlog conversion stays smooth; FlackTek integrates cleanly; commercial space orders convert without program deferrals; labor/materials costs stay contained.

Flag 1 — Valuation prices in flawless execution (HIGH severity) At approximately 95x trailing GAAP earnings (EDGAR-confirmed: $1,160.4M market cap / $12.23M FY2025 net income) and ~44x FY2026 guidance EBITDA, GHM trades at a multiple that has historically been reserved for businesses sustaining 20%+ EBITDA margins for multiple years. Graham has never achieved 13% adjusted EBITDA in its EDGAR history. The consensus analyst price target of $81.67 (late May 2026, 5 analysts) was 21% below the then-current price — the sell side had not modeled the premium-multiple expansion the market was implying. Ducommun trades at comparable scale with 17.5% EBITDA margins and a fraction of GHM's multiple.

Flag 2 — Proven history of violent revenue and margin collapses (HIGH severity) Revenue peaked at $135.2M (FY2015), then collapsed to $77.5M (FY2018) — a 43% decline in three years (EDGAR). Operating income turned deeply negative in FY2022 (-$11.3M, EDGAR). Net income was negative in both FY2018 (-$9.8M) and FY2022 (-$8.8M), and was barely breakeven at $367K in FY2023. The current ramp is real but the business has demonstrated it can destroy three-to-five years of earnings in a single order-lumpiness cycle. Investors paying 95x the first materially positive earnings year since FY2015 are pricing in a regime change the company has not yet demonstrated at scale.

Flag 3 — Defense concentration in a single appropriations chain (MEDIUM-HIGH severity) Approximately 85% of the $515.6M backlog is defense-related, and essentially all defense revenue flows through US Navy nuclear submarine programs via General Dynamics/Electric Boat and Huntington Ingalls. This is concentration within a single service branch's nuclear shipbuilding account. Columbia and Virginia programs are strategic deterrent assets and among the most protected items in the defense budget, but they still require annual appropriations. A continuing resolution, sequestration, or DOGE-driven budget cut would cause order-timing delays — not cancellations — but delays are sufficient to miss quarterly expectations and compress a 95x multiple.

Flag 4 — Margin target is unproven at scale (MEDIUM-HIGH severity) FY2025 GAAP operating margin was 7.2%; adjusted EBITDA margin nine-month FY2026 was approximately 10.8%. The FY2027 low-to-mid-teen target requires an additional 200-500 bps of expansion. Graham has never sustained 13%+ EBITDA margins in its EDGAR history. Peers Ducommun (17.5%) and Curtiss-Wright (16.6%) are already at or beyond GHM's target, having achieved it over multiple cycles — GHM is attempting to close that gap in a single fiscal year while simultaneously integrating a new acquisition.

Flag 5 — FlackTek integration and earn-out overhang (MEDIUM severity) The $35M FlackTek acquisition (closed January 23, 2026, 8-K confirmed) added a third manufacturing platform with up to $25M in earn-out payments over four years starting FY2027. If FlackTek hits its targets, Graham pays an additional $25M — a real cash obligation relative to $21.6M of year-end FY2025 cash. If FlackTek underperforms, management bandwidth has been consumed for a below-target return. The T. Rowe Price $50M timing overlap is not coincidental — the equity raise was needed to fund the acquisition and maintain balance sheet credibility.

Flag 6 — OCF divergence despite earnings improvement (LOW-MEDIUM severity) OCF fell from $28.1M (FY2024) to $24.3M (FY2025) even as operating income more than doubled from $6.9M to $15.2M. Working capital consumption is absorbing the operating leverage gains — consistent with a defense contractor ramping production and accumulating inventory/unbilled receivables. As the Columbia-class ramp accelerates, this drag could persist or worsen before backlog conversion flows to cash.

Flag 7 — Short interest rising +29.6% MoM ahead of June 8 earnings (LOW severity) Short interest was 403,906 shares as of May 15, 2026, up 29.6% month-over-month, at 3.68% of float with 2.1 days to cover. Not a crowded short — no squeeze protection. The re-accumulation of shorts immediately before earnings suggests institutional skepticism about Q4 delivery or valuation.

• The bear case's stated P/E of "74x" in its opening sentence is internally contradicted by the bear's own calculation of 94.8x in the same paragraph. The fact-sheet-confirmed figure is approximately 95x ($1,160.4M / $12.23M). The lower figure was unsupported.
• The bull's thesis paragraph states "17% incremental operating margin" but its own key-point calculation demonstrates 30.4% incremental margin ($26.5M recovery / $87.1M revenue increment, both EDGAR-confirmed). The 30% figure is arithmetically correct; the 17% figure in the thesis is unsupported.
• OCF has been consistently positive and meaningful since FY2023 after two years of negative OCF (FY2021-FY2022), but the FY2025 OCF decline is a genuine yellow flag given the simultaneous earnings improvement.
• The FlackTek earn-out is correctly characterized by the bear as a payment obligation from Graham's perspective, not a catalyst for existing shareholders — it is a future cash outflow if FlackTek performs, not an earnings contributor.
• Commercial space cryogenics is differentiated from defense by its lack of sole-source protection. Commercial space customers are schedule-volatile, and Graham's position as an independent supplier — while differentiated — does not carry the same switching-cost moat as Navy program content.
• T. Rowe Price's $83.36 entry price is approximately 20% below the current $104.15, meaning a sophisticated multi-year institutional buyer saw margin of safety at a price significantly below where the stock now trades.

Graham Corporation is a genuine, high-quality niche business with real program moats, a verified backlog, and an improving earnings trajectory that has not yet been fully discounted by thin sell-side coverage. The defense thesis is solid: Virginia-class and Columbia-class program anchors are among the most durable and politically protected in the US defense budget, and Graham's sole-source positions on specific sub-systems are protected by years of re-qualification lead time.

The problem is entirely about price. At approximately 95x trailing earnings and ~44x EV/EBITDA, the stock prices in not just the backlog conversion but a sustained mid-teen EBITDA margin regime that the company has never demonstrated in its EDGAR history. The business has twice collapsed to deeply negative earnings within the past decade, both times because defense order timing became lumpy — the same mechanism that created the current ramp. The consensus analyst price target ($81.67) was 21% below the current price even before the June 8 earnings call, and the bull's own fair-value math on FY2027 execution at a 25x EV/EBITDA premium peer multiple yields $80-90/share.

The risk/reward setup is asymmetric: the downside (margin target misses, Q4 disappointment, budget delay) compresses a 95x multiple substantially with no deterioration in the fundamental business; the upside (FY2027 targets hit, second tranche of space orders, analyst re-rating) offers 10-15% at current price with 30-35x EV/EBITDA required to hold. The T. Rowe Price $83.36 entry provides a useful reference: a sophisticated, long-only institution paid 20% less than the current price and considered that a fair entry point. The June 8 earnings call is the nearest discrete event that could resolve — in either direction — the margin trajectory question that underlies the valuation.

Research, not investment advice. Figures sourced from SEC filings and public data; verify before acting.