Espey Mfg. & Electronics Corp. (NYSE American: ESP)

Deep Dive Research Report - August 23, 2026

A note on reporting cadence and what the most recent period is

Espey's fiscal year ends June 30 and the company reports quarterly. Working forward from the FY2025 pattern (the FY2025 10-K was filed and the year-end release issued on September 16, 2025), the expected date for the FY2026 fourth-quarter and full-year release is mid-September 2026 - and there is positive evidence for that: the scheduled next-earnings date carried by data vendors is September 15, 2026, a future date (MarketBeat earnings calendar). The June 30, 2026 period is therefore not yet due, not merely unfound.

The most recent released period is Q3 FY2026, the quarter ended March 31, 2026, announced May 12, 2026 and filed on Form 10-Q the same day (Espey press release; 10-Q, esp-20260331).

On the six earnings calls. Espey does not hold earnings conference calls. It is a roughly 150-person company with no sell-side sponsorship; its quarterly disclosure consists of a GlobeNewswire release carrying a short CEO statement, plus the 10-Q or 10-K. No dial-in, webcast, or replay is announced in any of the releases reviewed, and no transcript exists on any aggregator. This is a real gap and I am flagging it rather than papering over it. For Sections 7 and 9 I have substituted the last six reporting periods - each results release plus its accompanying MD&A - and cited them by date exactly as I would cite a call:

1. What the company does

Espey makes the boxes that turn whatever electricity a military platform happens to have into whatever electricity a piece of equipment on that platform actually needs - and it makes them so that they keep working when a mine goes off next to the hull.

That is the whole business, and the second half of the sentence is the part that matters. Anyone can build a power supply. Very few companies can build one that survives MIL-S-901D Grade A hammer-blow shock testing, MIL-STD-167-1 vibration, salt fog, temperature extremes and the electromagnetic hygiene requirements of a submarine, and then document every step of how it was built so that a Navy program office can qualify it once and buy it for twenty years. Espey's products are power supplies, DC-DC and DC-AC converters, EMI filters, power transformers, inductors and other magnetic components, power distribution panels, solid-state circuit breakers, uninterruptible power systems, antennas and high-power radar transmitter assemblies (Espey products page). The applications named in its own filings run across AC and DC locomotives, shipboard power, shipboard radar, airborne power, ground-based radar and ground mobile power.

The company was incorporated in 1928 according to its SEC-filing lineage and corporate profiles, while Espey's own website dates the operating business "since 1930" (About Espey). Both dates circulate; I have not found a filing that reconciles them, so I am stating the discrepancy rather than choosing. What is not in dispute is that the business has spent the better part of a century in the same place - 233 Ballston Avenue, Saratoga Springs, New York - doing variations of the same thing. Its history page lists work on the C-130, the E-2C Hawkeye, the Littoral Combat Ship, the DDG-1000, early-warning radar high-voltage supplies, electronic warfare gear, missile guidance electronics, locomotive power supplies, and submarine-grade low-noise transformers (Espey history).

The value proposition, stated plainly

A prime contractor building a submarine or a radar has a problem: it needs a piece of power hardware that is electrically unremarkable but environmentally brutal, that has to be qualified to naval standards, that has to be sourced domestically, and that will be needed in low annual volumes for the next two or three decades. That is a terrible business for a large electronics company - the volumes are too small to matter and the qualification cost is high - and a terrible business for a small job shop, which cannot carry the quality system or the test lab.

Espey occupies exactly that gap. It is small enough to care about a 40-unit-a-year program and large enough to hold AS9100 certification, run its own environmental qualification lab, fabricate its own metalwork, populate its own boards, and hold a DoD-eligible contractor position (CAGE code 20950). Its own framing is that it operates "under large-business quality disciplines despite its small-business status" (About Espey). For a US Navy program office trying to qualify a domestic source for a transformer that goes into a ballistic-missile submarine, that combination is rare.

What it actually looks like for a customer

Take the December 2024 award. General Dynamics Electric Boat and the Navy needed electric power distribution panels for the Columbia class ballistic missile submarine. Espey had already designed those panels for the program. The award - $29.5 million when fully funded - was for manufacturing the next four of twelve planned ships, with deliveries running through calendar 2030 (Espey press release, December 23, 2024).

Then in April 2025, the same customer relationship produced a second award: $19.8 million when fully funded, for electrical power transformers for both the Virginia and Columbia classes - again, components "previously designed and supplied by Espey" for the Virginia class and the initial Columbia boats (Espey press release, April 10, 2025).

That is the machine in one paragraph. Espey spends money and engineering years up front to design a qualified part into a long-lived platform. Once the part is in the drawing package and the ship class is in serial production, Espey is the incumbent, and each successive block buy of hulls generates a multi-year production order that arrives without a competitive re-fight. The two awards above added $49.4 million of multi-year work in a single fiscal year - against a total backlog that, at June 30, 2024, had been $97.2 million (FY2025 results release, September 16, 2025).

The uncomfortable other half of the model

The same design-in that creates the annuity also creates the risk. Espey works predominantly under fixed-price contracts, recognising revenue by the output method - units delivered or milestones reached (Q3 FY2026 10-Q). On a build-to-print production run of a mature part, fixed price is fine. On a fixed-price engineering development contract, where Espey is designing something new to a specification, a technical surprise or a schedule slip is absorbed entirely by Espey. The FY2025 10-K sized that exposure explicitly: engineering programs in the funded and unfunded backlog together aggregated approximately $13 million (FY2025 10-K summary). That is the single most under-appreciated line in Espey's filings, and Section 8 returns to it.

2. Business segments

Espey reports as a single operating segment. It has no subsidiaries, no divisions with separate P&Ls, and no acquisition history creating distinct entities. Everything is designed, built and tested in one building in Saratoga Springs.

That said, the company organises its go-to-market around five product families, and those families have genuinely different economics, competitive sets and capital requirements. Treating them as de facto segments is the only way to understand the business, so I do that here - with the explicit caveat that Espey publishes no revenue split across them, so no mix percentages exist to cite.

2.1 Power conversion

What it does. MIL-SPEC-compliant power supplies and converters spanning 3.3V to 45,000V and 20W to 300kW (Espey home page). The catalogue includes DC-DC converters, DC-AC and AC-AC converters, power conditioning assemblies, battery chargers, hot-swappable supplies, uninterruptible power supplies and high-voltage supplies for radar transmitters. End platforms named by Espey include avionics and weapons systems, howitzers, missile systems, targeting pods, and shipboard equipment (military power systems page).

The core capability. High-voltage design at the top of that range is a genuinely narrow skill. A 45,000V supply for a radar transmitter is not a scaled-up version of a 28V converter - insulation coordination, corona suppression, potting and creepage distances become the design. Espey has been building radar transmitter power supplies for decades, which is why the history page can list early-warning radar high-voltage supplies as a heritage line rather than a new venture.

Why it stands apart. Power conversion is the product family where Espey most directly meets branded merchant competition - Vicor, Astrodyne TDI, Behlman - because a converter is closer to being a catalogue item than a submarine transformer is. It is also the family with the most airborne and ground content, so it diversifies Espey away from naval programs.

Competitive position. Espey wins where the environmental spec is extreme, the volume is low and the customer wants a single qualified domestic source. It loses where a merchant COTS or MIL-COTS module from Vicor or Astrodyne will pass the spec at lower unit cost - which is increasingly often at the lower voltage and power end, as wide-bandgap semiconductors push merchant modules further into ruggedised territory.

2.2 Power magnetics

What it does. AC-conditioned transformers rated 12V to 45,000V and 20W to 3.8MW, plus inductors and related magnetic components (Espey home page). Naval distribution transformers and submarine low-noise transformers are the marquee applications.

The core capability. This is Espey's deepest moat and the hardest thing in the building to replicate. A submarine power transformer is not judged primarily on efficiency; it is judged on acoustic signature, shock survivability and magnetic stray field. A transformer that hums is a transformer that can be heard. Getting a large magnetic assembly to be simultaneously quiet, shock-hardened to MIL-S-901D Grade A, and manufacturable is process knowledge accumulated over decades - core stacking, winding tension, impregnation, mounting isolation - and it does not transfer easily to a new entrant even with the drawings.

Why it stands apart, and why the Navy paid to expand it. Magnetics is capital-intensive and capacity-constrained in a way power conversion is not, and the US Navy treated Espey's magnetics capacity as a national industrial-base problem rather than a supplier problem. In February 2023 the Navy awarded Espey a $7.4 million grant under the Surface Combatant Industrial Base Development Initiative to build a dedicated facility; a further $3.4 million followed for capital equipment and facility upgrades, with project completion expected in 2026 (Espey funding award release). The resulting 24,000 square foot Magnetics Center of Excellence completed construction in April 2025 and reached full production and testing operations in October 2025 (Espey press release, October 27, 2025).

Read that carefully: the customer funded roughly $10.8 million of the supplier's capacity expansion. That is not a normal commercial relationship. It is the behaviour of a buyer who has concluded there is no substitute.

Competitive position. Strongest of the five families. The relevant competitors are not merchant power companies but a short list of qualified naval magnetics shops, and the qualification barrier is the whole game.

2.3 Build-to-print manufacturing services

What it does. Espey manufactures to a customer's existing drawing package - no design content, just execution. It also sells discrete services: metal fabrication, painting, environmental testing, design studies, and development of automatic test equipment (FY2025 10-K, Item 1).

The core capability. Vertical integration. Espey populates boards, fabricates metalwork, paints, and performs mechanical, electrical and environmental qualification testing in-house (About Espey). For a build-to-print job that is decisive: the customer hands over one drawing package and receives a finished, tested assembly rather than managing five vendors.

Why it stands apart. Build-to-print carries no design risk and no fixed-price engineering exposure - and correspondingly less pricing power. Management explicitly attributed part of the FY2025 growth to "several large multi-year contracts and increases in build to print sales" (FY2025 10-K MD&A). It is the load-balancing family: it fills the factory between engineering-heavy programs.

Competitive position. Weakest structural position of the five, because a drawing package is portable. Espey competes here on schedule reliability, security clearance posture and the fact that a customer already qualified on Espey's line does not want to re-qualify elsewhere.

2.4 Integrated power systems

What it does. Espey markets integrated systems for energy storage and directed energy weapons applications (Espey home page). This is the newest and least-proven family.

The core capability. Directed energy is a power problem before it is a laser problem: the pulse-forming, storage and conditioning chain that lets a platform dump megawatt-class energy in milliseconds without browning out the ship. Espey's existing skill in high-voltage conversion and large magnetics maps onto that chain directly.

Why it stands apart. It is the optionality line - the place where a decades-old transformer company gets to participate in a new mission area using the same physics. It is also almost certainly where the fixed-price engineering exposure lives.

Competitive position. Unproven. Espey has published no directed-energy program wins, and the credible read is that this is a development and positioning effort rather than a revenue line today.

2.5 Power distribution and solid-state circuit breakers

What it does. Power distribution panels and solid-state protection and distribution equipment. The Columbia class distribution panels behind the $29.5 million December 2024 award sit here.

The core capability. Distribution panels are where power hardware becomes shipbuilding: physical form factor, shock mounting, cable entry, arc containment and the ability to survive installation in a hull. The design is inseparable from the ship.

Why it stands apart. This family has the longest program tails of any. A panel designed into the Columbia class has twelve hulls of demand ahead of it stretching well past 2040.

Competitive position. Effectively incumbent-locked on the programs Espey already holds; effectively locked out of the ones it does not.

Summary comparison

3. Products and business detail

The catalogue, product by product

High-voltage power supplies. Up to 45,000V. These feed radar transmitters and electronic warfare equipment. The engineering challenge is insulation and corona management in a sealed, shock-mounted enclosure that has to work at altitude, where reduced air density makes arcing dramatically easier. Espey's history page cites high-voltage supplies for early-warning radar as a heritage product line.

DC-DC, DC-AC and AC-AC converters. The workhorse family. On an aircraft, the platform supplies 28VDC or 115VAC 400Hz; a sensor may need multiple isolated rails at tight regulation. On a ship, the bus is 440VAC 60Hz and the equipment may want anything. These converters are the translation layer.

Uninterruptible and hot-swappable power supplies. Combat systems cannot tolerate a power interruption during a bus transfer. Hot-swap capability means a failed module can be replaced without taking the equipment offline - a maintainability requirement that shapes the mechanical design as much as the electrical one.

Power transformers and magnetic components. 12V to 45,000V, 20W to 3.8MW. The 3.8MW top end is ship-service scale hardware, not electronics-box scale. Submarine low-noise transformers are the technically hardest item Espey makes.

Power distribution equipment and solid-state circuit breakers. Panels that take ship power and split it to loads with protection and monitoring. Solid-state breakers replace mechanical contacts with semiconductors, giving faster trip response, arc-free operation and the ability to report status to a ship's control system - increasingly a requirement on new-build hulls.

EMI/power filters. Every one of the above has to not pollute the platform's electromagnetic environment, and has to tolerate what the platform throws at it. On a submarine this is a signature issue, not just a compliance issue.

Battery chargers, including aircraft support chargers. Ground support equipment, less glamorous, steady demand.

Antennas and high-power radar systems; digital fire control electronics. Legacy but current product areas listed on the military systems page, sitting outside the pure power franchise.

The standards that define the product

Espey's own product literature cites compliance with MIL-T-15108C (Navy) for transformers, MIL-S-901D Grade A high-impact shock, MIL-STD-167-1 vibration, Army Ground Mobile specifications, and ABS-NVR, IEC and IEEE standards (military power systems page). The company holds ISO AS9100 certification - the aerospace quality standard - and a DoD-eligible contractor position under CAGE code 20950.

MIL-S-901D Grade A deserves explanation because it is the single most consequential barrier in this business. It is a live explosive shock test: the article is mounted on a floating shock platform and explosive charges are detonated at decreasing standoff distances. There is no simulation shortcut and no partial credit. A design either survives or it is redesigned and retested, and each test cycle costs real money and months. That is why naval power hardware qualification is a multi-year, multi-million-dollar undertaking, why programs are reluctant to re-open a source once qualified, and why Espey's installed base is worth more than its order book.

Manufacturing

Everything happens in one place. Espey's website describes a 150,000+ square foot facility at 233 Ballston Avenue in Saratoga Springs where all design, manufacturing and testing are performed (About Espey); the FY2025 10-K describes the property as 174,000 square feet (FY2025 10-K summary) - the difference is presumably manufacturing space versus total including the new expansion and support areas.

In-house capability spans circuit board population, sheet metal fabrication, painting, magnetics winding and impregnation, mechanical assembly, and a qualification lab that performs mechanical, electrical and environmental testing. Espey also develops its own automatic test equipment - a detail worth pausing on, because on a low-volume program the ATE is often as much engineering as the product, and outsourcing it means handing a third party your acceptance criteria.

The Magnetics Center of Excellence is the most important physical change to the business in decades. 24,000 square feet, construction complete April 2025, full production and testing operations from October 2025, funded by a $7.4 million Navy grant received in FY2023 under the Surface Combatant Industrial Base Development Initiative, with a further $3.4 million for capital equipment and facility upgrades completing in 2026. The stated purpose is development, fabrication and MIL-STD testing of transformers, inductors and related components, with enhanced testing capability and higher power capacity than the legacy area (Espey press release, October 27, 2025).

CEO David O'Neil framed it this way in that release:

"This new 24,000-square-foot expansion further supports Espey's market leadership in advanced magnetics systems development..."

Workforce

152 employees as of August 31, 2025. Approximately 34% are represented by the International Brotherhood of Electrical Workers under a collective bargaining agreement expiring June 30, 2028; the 10-K characterises relations as good (FY2025 10-K). Current open roles span drafting, test, supply chain and quality control (Espey careers).

For a business whose constraint is skilled magnetics winders and test technicians in a small upstate New York labour market, 152 people is the real capacity number. A new building does not produce transformers; people in the new building do.

Geographies and exports

Espey is overwhelmingly domestic. Export shipments were $3,124,820 in FY2025, up on the prior year primarily because of one large power supply contract (FY2025 10-K). The 10-K describes marketing through Espey's own direct sales organisation to large industrial manufacturers and defense companies, the US government, foreign governments and major foreign electronic equipment companies. There is no distributor network and no overseas facility. Domestic sourcing is a feature, not a limitation: much of the demand exists precisely because the Navy wants the part made in the United States.

Milestones that changed the business

• February 2023 - $7.4 million Navy grant under the Surface Combatant Industrial Base Development Initiative to construct a new manufacturing facility.
• December 23, 2024 - $29.5 million Columbia class distribution panel award from General Dynamics Electric Boat and the Navy, deliveries through calendar 2030.
• April 10, 2025 - $19.8 million Virginia and Columbia class transformer award. Combined with the December award, $49.4 million of multi-year work booked inside FY2025.
• April 2025 - Magnetics Center of Excellence construction complete.
• June 30, 2025 - Backlog of $139.7 million, described by the company as a record, against $97.2 million a year earlier. FY2025 new orders of $86.4 million against $52.4 million in FY2024 (FY2025 results release, September 16, 2025).
• September 30, 2025 - Backlog peaks at $141.1 million (Q1 FY2026 release, November 12, 2025).
• October 27, 2025 - Magnetics Center of Excellence enters full production and testing.
• Additional $3.4 million Navy capital-improvement award, project completion expected 2026.

4. Customers

Who buys

Espey's customer base is a short list of large defense primes and the US Government. Its filings and marketing name work on display and radar applications at Boeing, Raytheon, Lockheed Martin, Northrop Grumman and General Dynamics, among others (FY2025 10-K, Item 1). The most consequential named relationship is with General Dynamics Electric Boat, which together with the Navy selected Espey for both submarine awards.

Espey does not name customers against revenue percentages in its concentration disclosures - it discloses unnamed "significant customers." Combining the two disclosures is the useful exercise, and it is done below.

The concentration picture, and how it is moving

Sources: FY2025 10-K; Q3 FY2026 10-Q.

Two readings are possible and both are partly right. The nine-month FY2026 number shows genuine improvement - four customers at 61% is a materially lighter concentration than six at 74%. But the single-quarter Q3 figure, five at 73% with a 20% top customer, shows how quickly a shipment-timing swing re-concentrates the book. The honest conclusion: concentration is structurally high and will remain so, because the customer set is structurally small. Management has stated it is targeting marketing at a broader set of defense contractors and government agencies specifically to reduce concentration (FY2025 10-K MD&A).

Who inside the customer decides, and on what

For a new design-in, the decision sits with a prime's engineering and supply-chain organisation, usually with the government program office in the room, because a qualified naval source is a program asset rather than a purchasing convenience. Criteria in rough order: does the supplier hold or can it obtain the required qualification; can it be a sole domestic source; does it have documented process control (AS9100); does it have the test capability; and only then, price. Sales cycles for this are measured in years, spanning proposal, development contract, qualification testing, first article and production release.

For a follow-on production buy on a part Espey already owns, the decision is close to automatic. The $29.5 million and $19.8 million awards are both explicitly for components Espey had previously designed for those programs. That is not a competitive win; it is the exercise of an option the customer created years earlier.

For build-to-print, the buyer is a prime's procurement organisation, criteria are schedule, quality history and price, and cycles are short. This is the contestable part of the book.

Switching costs

They are among the highest in industrial manufacturing, and they are not contractual - they are physical and regulatory.

To replace Espey on a qualified naval part, a customer must find a domestic supplier willing to invest in the capability, transfer the technical data package, have the new source build first articles, and then re-run the qualification - which for shipboard equipment means re-running MIL-S-901D shock, MIL-STD-167-1 vibration and the full environmental suite. That is a multi-year, multi-million-dollar exercise on a part that may only cost a few hundred thousand dollars a year. For a submarine program already fighting to increase build rate, deliberately introducing that risk is close to unthinkable.

The corollary is that switching costs cut both ways: they protect Espey's installed base and they lock Espey out of a competitor's installed base just as firmly. Espey's addressable growth is new designs and new ship classes, not share gains on existing parts.

Contract structure and revenue predictability

Espey works predominantly under fixed-price contracts, with revenue recognised by the output method - appraisal of results achieved, milestones reached, or units delivered. In the nine months to March 31, 2026, $28.70 million was recognised on units delivered and $3.95 million on milestones achieved (Q3 FY2026 10-Q).

The funding structure is where the real predictability lives, and it improved sharply. At June 30, 2025, backlog was $139.7 million of which $106.6 million was funded and $33.0 million unfunded. By March 31, 2026, backlog was $137.1 million and "fully funded, with the exception of approximately $14.5 million," which management believes will receive future appropriations based on program status. Unfunded backlog more than halved in nine months as appropriations caught up with awards.

Backlog phasing at March 31, 2026: 11% in FY2026, 38% in FY2027, 22% in FY2028, 29% thereafter. Note what that says - the current year is nearly finished shipping out of backlog, and the peak conversion year is FY2027. Espey also carries $33.49 million of contract liabilities, meaning customer prepayments, which is a substantial working-capital gift for a company this size and another signal of how much the customer wants the capacity to exist.

5. Competitive landscape

The structure of the market

There is no single "Espey market." There are three overlapping arenas with different competitors, and Espey's position varies from near-unassailable to genuinely exposed depending on which one you are looking at.

Arena 1: Qualified naval magnetics and distribution hardware. This is a small, qualification-gated arena where the competitor set is a handful of firms that have already survived shock testing on a naval program. Espey's own filings describe competition coming from "divisions of large electronics companies as well as smaller manufacturers," competing on "price, product performance, the experience of the particular company," and historical relationships - and note that competitors sometimes accept lower margins strategically (FY2025 10-K). The Navy funding Espey's expansion is the clearest available evidence that this arena is supply-constrained rather than demand-constrained.

Arena 2: Ruggedised merchant power conversion. This is a real, branded, commercially aggressive market. Vicor, Astrodyne TDI and Behlman all sell MIL-COTS and VPX-format modules aimed at exactly the airborne, ground and shipboard applications Espey also serves. The competitive dynamic here is unfavourable to Espey over time: wide-bandgap semiconductors (SiC, GaN) let merchant vendors push power density and thermal performance further into ruggedised territory each product generation, and a customer who can buy a qualified module off a datasheet will not commission a custom design.

Arena 3: Build-to-print contract manufacturing. Fragmented, price-competitive, low barriers beyond quality certification and clearances. Espey competes on vertical integration and delivery record, not on structural advantage.

Named competitors

Market cap sources: companiesmarketcap - Vicor; companiesmarketcap - Curtiss-Wright; stockanalysis - Orbit International. Competitor identification from Espey 10-K competition discussion and third-party competitor profiles.

A further category of competitor never appears in a comparison table: the prime's own in-house capability. General Dynamics, Northrop Grumman, L3Harris and Leonardo DRS all have internal power electronics groups. Their make-versus-buy decision is the largest single competitive variable Espey faces and it is entirely outside its control.

Barriers to entry - how high, honestly

High in Arena 1, and this is not a marketing claim. The barrier is a stack: capital for a magnetics facility and a qualification lab; the trained winders and test technicians; AS9100 and DFARS-compliant cybersecurity posture; a CAGE code and clearances; and then a multi-year, live-explosive qualification campaign per part. That last item is what makes the barrier durable rather than merely expensive - it costs time that cannot be bought.

Low to moderate in Arenas 2 and 3. A capable contract manufacturer with AS9100 can win build-to-print work. A well-funded semiconductor-led power company can win merchant conversion sockets.

Market share

No credible published share data exists for a niche this specific, and I am not going to invent one. The useful framing is that Espey's total annual output is small relative to the multi-billion-dollar military power supply market described in Section 6, so its market-level share is de minimis. Its share is meaningful only program by program, where on the parts it owns it is frequently the sole qualified source. That is the correct mental model: Espey does not have market share, it has a portfolio of monopolies on individual part numbers.

Structural shifts underway

• Submarine industrial base consolidation and government-directed investment. The Navy is directly funding supplier capacity - including Espey's. This favours incumbents with existing qualifications and disfavours anyone trying to enter now, because the government is buying capacity from firms it has already qualified.
• Wide-bandgap semiconductors. SiC and GaN raise switching frequencies and shrink magnetics. Over a long horizon this is a genuine threat to a magnetics-heavy franchise, though naval acoustic and shock requirements slow adoption far more than in commercial aerospace.
• Solid-state distribution. Espey is on the right side of this one, selling solid-state circuit breakers into new-build hulls.
• Domestic-content pressure. Every tightening of US sourcing requirements is a tailwind for a company that has never manufactured anywhere but Saratoga Springs.

Where Espey is strong and where it is exposed

Strong: submarine magnetics and distribution hardware, where qualification and customer-funded capacity create an incumbency that money alone cannot dislodge; high-voltage radar supplies, where the design skill is scarce; and vertical integration, which shortens schedules.

Exposed: the merchant conversion end of the catalogue, where technology moves faster than Espey's R&D budget can follow; build-to-print, where the drawing package is portable; the prime make-versus-buy decision; and above all the fact that a company of roughly 150 people cannot defend seven product families equally. Espey's competitive position is deep and narrow, and the narrowness is not a fixable condition.

6. Industry

What drives demand

Demand for Espey's products is a derivative of two things: how many US military platforms are being built, and how much electrical power each platform needs. Both are moving in Espey's favour simultaneously, which is unusual.

The platform-count driver is the dominant one and it is concentrated in submarines. The FY2026 budget provided roughly $10.5 billion for Columbia class procurement, $1.9 billion for other Columbia purposes, and $1.35 billion specifically for industrial-base growth on that program (Overt Defense, April 1, 2026). On July 30, 2026 the Navy awarded $76.6 billion in contracts for Virginia and Columbia class submarines (Defense News; Breaking Defense). Looking forward, the FY2027 President's Budget Request contains $65.8 billion for naval shipbuilding, a 46% increase over FY2026 and 123% over FY2025, with $124.9 billion allocated to submarines across FY2027-FY2031, plus $6.2 billion to grow submarine industrial base production capacity and $7.2 billion for nuclear shipbuilder productivity (Westside Construction Group analysis of FY2026-27 shipbuilding investment).

The power-density driver is quieter but structural. Armed forces are integrating more electronics into every platform - sensors, processors, datalinks, electronic warfare, AI modules - which raises the requirement for power density and efficiency in every box (Fortune Business Insights). A modern destroyer or submarine simply has more things that need conditioned power than its predecessor did.

Industry size and growth

• The generation military power supply market is projected to grow from approximately US$12.84 billion in 2026 to US$23.09 billion by 2034, a CAGR of about 7.6% (Fortune Business Insights).
• The next-generation military power supply market was valued at approximately US$15.29 billion in 2026, projected to US$20.34 billion by 2030, a CAGR of about 7.4% (Research and Markets, March 2026).

These figures are for the global addressable category, not for anything Espey could realistically capture. Their value here is directional: the category grows mid-to-high single digits, while Espey's specific naval-submarine sub-segment is growing far faster than that because of the shipbuilding budget trajectory above.

Where Espey sits in the supply chain

Third tier, sometimes second. The Navy contracts with General Dynamics Electric Boat and Huntington Ingalls Newport News; those yards contract with system integrators and component suppliers; Espey supplies qualified power components either to the yard directly or into a prime's system. Espey never sees the platform contract. It sees a purchase order whose timing is downstream of appropriations, yard schedule and long-lead material availability - which is exactly why quarterly shipment timing at Espey is lumpy in ways that have nothing to do with underlying demand.

Import substitution

This is close to inverted for Espey. There is no meaningful import competition in qualified naval power hardware, because the requirement is domestic by policy and by security classification. The relevant dynamic is not import substitution but domestic capacity creation - the government paying to expand US supplier capability because too little exists. Espey's $7.4 million plus $3.4 million of Navy facility funding is a direct instance of that policy.

The import exposure runs the other way, on inputs: the FY2025 10-K flags uncertainty around tariffs on steel and aluminium imports, which are raw materials for transformer cores and enclosures.

Regulatory environment

Four regimes shape the market:

1. Military specifications - MIL-T-15108C, MIL-S-901D, MIL-STD-167-1 and the rest. These are the qualification barrier described above.
2. AS9100 / ISO quality certification, without which a prime will not place work.
3. DFARS cybersecurity obligations, which the FY2025 10-K names as a compliance cost and a risk. For a 152-person company, meeting defense cybersecurity requirements is a genuine fixed-cost burden.
4. Appropriations. Espey's backlog can be awarded and still not be funded. The $14.5 million unfunded portion at March 31, 2026 exists entirely because Congress has not yet appropriated against it.

Cyclicality

Defense electronics is not economically cyclical in the ordinary sense; it is politically cyclical. The relevant cycle is the budget cycle and the shipbuilding plan, both of which currently point strongly upward. What Espey does experience is a shipment-timing cycle with a period of a quarter or two, driven by customer schedule shifts, government shutdown risk and component lead times - and that is the source of essentially every quarter-to-quarter revenue swing in the last two years.

Tailwinds and headwinds

Tailwinds: an unprecedented submarine build-rate push with money attached; government-funded supplier capacity expansion; domestic-content preference; rising electrical load per platform; directed energy and electrification creating new power-hardware categories.

Headwinds: supply chain lead times, with some components approaching one-year delays as of the FY2025 10-K; skilled labour scarcity in specialised manufacturing; steel and aluminium tariff uncertainty; cybersecurity compliance cost falling disproportionately on small suppliers; and the perennial risk that a continuing resolution or shutdown stalls the appropriations that convert awards into funded orders.

7. Growth triggers

Source note. Espey holds no earnings conference calls. The triggers below are drawn from the six most recent quarterly reporting events - each results release and its accompanying MD&A - plus specific 8-K contract announcements, and each is cited to its date. Nothing here is inference; it is all what management stated.

• Magnetics Center of Excellence ramp to full output. 24,000 square feet, construction complete April 2025, full-scale production and testing operations underway from October 2025. Adds testing capability and power capacity for transformers, inductors and related magnetics. (Espey press release, October 27, 2025; repeated context in Q1 FY2026 release, November 12, 2025)

"This new 24,000-square-foot expansion further supports Espey's market leadership in advanced magnetics systems development..." - David O'Neil, President and CEO, October 27, 2025

• A second Navy capital award completing in 2026. $3.4 million for capital equipment and facility upgrades, on top of the earlier $7.4 million facility grant, with project completion expected in 2026. (Espey funding award announcement)

• Columbia class distribution panel production running to 2030. $29.5 million when fully funded, covering the next four of twelve planned ships, deliveries expected through calendar year 2030. (Announced December 23, 2024; carried in the backlog reported at every subsequent period)

• Virginia and Columbia class transformer production. $19.8 million when fully funded, for components Espey had previously designed and supplied for both classes. (Announced April 10, 2025; referenced by management in the Q3 FY2025 release, May 12, 2025)

On Q3 FY2025, management stated that backlog growth was "a direct result of an exceptional year for new orders, including the previously announced new contract for $19.8 million." - Q3 FY2025 release, May 12, 2025

• Backlog conversion peaking in FY2027. At March 31, 2026, backlog phasing was 11% FY2026, 38% FY2027, 22% FY2028, 29% thereafter. FY2027 is the heaviest scheduled shipping year in the current book. (Q3 FY2026 10-Q, May 12, 2026)

• Unfunded backlog converting to funded. Unfunded backlog fell from $33.0 million at June 30, 2025 to approximately $14.5 million at March 31, 2026, with management stating it believes the remainder will receive future appropriations based on program status. (FY2025 10-K, September 16, 2025; Q3 FY2026 10-Q, May 12, 2026)

• A quantified bid pipeline. Management disclosed approximately $152.5 million of outstanding bid opportunities across repeat and new programs. (Q3 FY2026 10-Q, May 12, 2026) - This is the single most important forward number Espey publishes, because FY2026 order intake has been weak and this is the pipeline that has to replace it.

• Explicit full-year FY2026 guidance. Management stated it expects higher revenues in FY2026 than FY2025, and that FY2026 net income will exceed FY2025 net income. (Q3 FY2026 10-Q, May 12, 2026)

"With a strong quarter behind us, we continue accelerating into the remainder of the year with confidence" in delivering "excellent full-year results." - David O'Neil, Q3 FY2026 release, May 12, 2026

• Second-half shipment catch-up on delayed deliverables. Management attributed H1 FY2026 top-line softness to external factors that shifted the timing of certain deliverables, and committed to executing those milestones in the second half. (Q2 FY2026 release, February 10, 2026; the same timing-shift explanation appeared in the Q1 FY2026 release, November 12, 2025)

"We delivered solid margins, made meaningful progress across key programs, and continue to see encouraging trends as we move through the year." - David O'Neil, Q2 FY2026 release, February 10, 2026

• Deliberate customer diversification effort. Management stated it is targeting marketing at defense contractors and government agencies specifically to reduce customer concentration. (FY2025 10-K MD&A, September 16, 2025) - This has partially delivered already: significant-customer concentration for the nine months to March 2026 was four customers at 61%, against six at 74% for FY2025.

• Cost and efficiency programme. Management repeatedly identified labour efficiencies, negotiated material savings and resource management as the levers behind gross profit improvement. (Q1 FY2026 release, November 12, 2025; reiterated Q2 FY2026, February 10, 2026 and Q3 FY2026, May 12, 2026; originally flagged as "bottom-line cost saving initiatives" in the Q2 FY2025 release, February 12, 2025)

Trigger summary

8. Key risks

1. Order intake has collapsed against a record backlog, and the backlog is finite

Mechanism. Espey's FY2025 was an outlier: $86.4 million of new orders, driven by the $29.5 million and $19.8 million submarine awards. FY2026 has been the mirror image. New orders were $10.5 million in Q1, $16.3 million cumulative through H1, and $30.0 million cumulative through nine months - against $75.1 million in the first nine months of FY2025. Management stated plainly in the Q3 FY2026 10-Q that it expects full-year FY2026 orders to be lower than FY2025.

Backlog has already begun to roll over as a result: $141.1 million at September 30, 2025, $134.7 million at December 31, 2025, $137.1 million at March 31, 2026. Backlog declines when shipments exceed orders, and that is what has been happening. With 11% of backlog scheduled to ship in the remainder of FY2026, 38% in FY2027 and 22% in FY2028, the current book carries Espey comfortably for roughly two years - after which order intake has to recover or the factory runs out of work.

Calibration: high probability of continued soft intake in the near term, moderate consequence, with a long fuse. The mitigant is the $152.5 million bid pipeline management disclosed, plus the Navy shipbuilding budget trajectory. But a bid pipeline is not an order book, and the gap between $86.4 million and roughly $40 million of annual intake is the single number to watch when the FY2026 10-K lands in September.

2. Fixed-price engineering contracts, where Espey eats the overrun

Mechanism. Espey works predominantly under fixed-price contracts. On a mature build-to-print part, fine. On an engineering development contract, a technical failure or schedule slip is absorbed entirely by Espey - and unlike a large prime, Espey has no portfolio across which to spread a bad program. The FY2025 10-K sized the exposure directly:

"Engineering programs in both the funded and unfunded portions of the current backlog aggregate $13 million." - FY2025 10-K

The Q3 FY2026 10-Q reiterated that fixed-price engineering contracts can create loss exposure if technical or schedule challenges arise.

Calibration: moderate probability, high consequence for a single reporting period. For a company of this size, one bad engineering program can dominate a year's results. Note also that the directed-energy and energy-storage work in Section 2.4 is precisely the kind of activity that generates fixed-price development contracts.

3. Customer concentration that cannot be structurally fixed

Mechanism. Six customers were 74% of FY2025 sales; five customers were 73% of Q3 FY2026 sales with a top customer at 20%; three customers were 51% of FY2025 receivables. Losing one relationship - or having one prime lose a program, or decide to make the part in-house - removes a large fraction of the business at once. The receivables concentration adds a second, less-discussed exposure: a single prime's payment behaviour materially affects Espey's working capital.

Calibration: low probability of an abrupt loss given qualification lock-in, high consequence if it happens. Management is working on it and has made real progress at the nine-month level, but the underlying customer universe is small and always will be.

4. Appropriations, continuing resolutions and shutdowns

Mechanism. Espey can hold an award and still not be able to build against it. Approximately $14.5 million of the March 31, 2026 backlog was unfunded, pending future appropriations. The Q3 FY2026 10-Q named government-related timing and shutdowns among its risk factors. A prolonged continuing resolution stalls new-start funding, and a shutdown stalls contracting officers, deliveries and payments.

Calibration: high probability of intermittent disruption, moderate consequence each time. This is the most likely explanation for the recurring "timing shift for certain deliverables" language across the FY2026 releases.

5. Supply chain lead times on components with no substitute

Mechanism. The FY2025 10-K flagged supply chain constraints with some components approaching one-year lead times. On a MIL-qualified assembly a component is not fungible: the part number is in the qualified configuration, and substituting it can require re-qualification. So Espey cannot simply buy an alternative when a supplier goes long - it waits, and the shipment slips.

Calibration: high probability, moderate ongoing drag. This directly explains lumpy quarterly shipments and is why backlog conversion timing keeps moving.

6. Skilled labour in a small upstate New York market

Mechanism. 152 employees, roughly a third IBEW-represented under an agreement expiring June 30, 2028. Magnetics winders, MIL-spec assemblers and qualification test technicians are not roles that can be hired quickly. The Magnetics Center of Excellence adds 24,000 square feet of capacity that only produces if it is staffed, and Espey is currently recruiting in drafting, test, supply chain and quality control. A labour shortfall converts a capacity investment into idle floor space.

Calibration: moderate probability, moderate consequence - and it is the specific mechanism by which the growth thesis could quietly fail to convert.

7. Wide-bandgap semiconductors eroding the magnetics content

Mechanism. SiC and GaN devices switch faster, which allows smaller magnetics for a given power level. Every generation of merchant power electronics shrinks the transformer. Over a decade-plus horizon this compresses the value of magnetics content in a power system and lowers the barrier for merchant vendors to enter applications that today require a custom Espey design.

Calibration: low probability of near-term damage, high probability of long-term erosion. Naval acoustic and shock requirements slow this considerably - a quiet submarine transformer is not a switching-frequency problem - but it is a structural direction of travel, not a cycle.

8. Tariffs on transformer raw materials

Mechanism. The FY2025 10-K states that tariff impacts on steel and aluminium imports remain uncertain. Electrical steel and aluminium are direct inputs to transformer cores and enclosures. Under a fixed-price contract signed before a tariff change, the cost increase has nowhere to go.

Calibration: moderate probability, low-to-moderate consequence.

9. DFARS cybersecurity compliance as a fixed cost on a small base

Mechanism. The FY2025 10-K names cybersecurity threats and DFARS compliance obligations as risks. Defense cybersecurity requirements do not scale down for small suppliers: the control set is the control set. For a 152-person company that is a disproportionate fixed cost, and a compliance failure could jeopardise eligibility to hold controlled technical data - which is existential rather than expensive.

Calibration: low probability, catastrophic consequence in the tail.

9. Walk the talk

The six reporting periods used: Q2 FY2025 (released February 12, 2025), Q3 FY2025 (May 12, 2025), Q4/FY2025 (September 16, 2025), Q1 FY2026 (November 12, 2025), Q2 FY2026 (February 10, 2026), Q3 FY2026 (May 12, 2026). The most recent is 103 days before today, which is as recent as it is possible to be - the FY2026 year-end release is not due until roughly September 15, 2026.

The structural caveat, stated once and honestly: Espey holds no earnings calls. There is no Q&A, no analyst pressing management on a missed number, no unscripted moment. Management's public record consists of two or three sentences per quarter in a press release plus the MD&A. That is a thinner evidentiary base than this section would normally work from, and it means Espey's management is easier to assess on outcomes than on candour under questioning, because nobody ever questions them. I have weighted outcomes accordingly.

Starting point: Q2 FY2025, February 12, 2025

The oldest of the six is the high-water mark of optimism. Management announced record quarterly sales and a record backlog of $120.1 million, noted the backlog included the previously announced $29.5 million Columbia class order, and reported six-month new orders of $46.9 million - already, at the halfway mark, close to the prior full-year record. The forward commitment was specific and modest:

Management is prioritising "bottom-line cost saving initiatives" to enhance future program margins. - Q2 FY2025 release, February 12, 2025

That is the promise to track. It is not a revenue promise; it is a margin-mechanics promise. Hold that thought.

Q3 FY2025, May 12, 2025

Backlog rose again, to $138.0 million. Nine-month new orders reached $75.1 million, against $27.8 million in the same period of FY2024. Management attributed the growth directly to order intake:

Backlog growth was "a direct result of an exceptional year for new orders, including the previously announced new contract for $19.8 million." - Q3 FY2025 release, May 12, 2025

Note what management did here: it named the specific contract responsible rather than describing the growth in general terms. That is a small thing, but it is the difference between an attributable claim and a mood.

Q4/FY2025, September 16, 2025

The year closed with backlog at $139.7 million against $97.2 million a year earlier, and full-year new orders of $86.4 million against $52.4 million. The company described the backlog as being at record levels. Gross margin improved to 28.9% from 27.5% - the first visible payoff on the cost-savings promise made in February.

And then management did something genuinely unusual for a company with no analyst coverage and no obligation to be candid: it guided down on profitability while guiding up on revenue. The FY2025 10-K stated that management expected higher fiscal 2026 revenues but a decline in net income per share "due to higher costs in the product mix."

That is a management team volunteering bad news nobody was going to ask about. It is the single most credible thing in this entire six-period record, and it sets up the most interesting question in the section: did the guide-down turn out to be right?

Q1 FY2026, November 12, 2025

It did not - in the company's favour. Sales came in below the prior-year quarter, but backlog reached its peak of $141.1 million and new orders were $10.5 million against $7.8 million. Management explained the sales shortfall without hedging:

"Despite softer sales, we saw significant improvement in gross profit, backlog levels, and net income results," with the sales decrease attributed to "a change in the timing of shipments for the year." - David O'Neil, Q1 FY2026 release, November 12, 2025

He then named the mechanism behind the margin improvement - "labour efficiencies, negotiating savings on materials, and effective resource management" - which is the same list from the February 2025 cost-savings promise, now delivering. Promise made, promise kept, three quarters later.

Q2 FY2026, February 10, 2026

The first genuinely uncomfortable quarter. Sales fell against the prior-year comparative and, more importantly, six-month new orders were $16.3 million against $46.9 million - a two-thirds collapse in intake. Backlog began to decline, to $134.7 million. Management's framing:

"We delivered solid margins, made meaningful progress across key programs, and continue to see encouraging trends as we move through the year." He attributed some top-line softness to external factors causing delivery timing shifts, expressing confidence in second-half execution. - David O'Neil, Q2 FY2026 release, February 10, 2026

Two observations. First, the release did disclose the order collapse in plain numbers, including the explanatory detail that the prior-year comparative contained $29.5 million of multi-year awards. It did not bury it. Second, the CEO's quoted commentary did not mention orders at all. The disclosure was complete; the narrative was selective. That is a real, if minor, mark against.

Q3 FY2026, May 12, 2026

The second-half execution promise was delivered. Quarterly sales rose against the prior-year quarter, gross margin reached 37.0% for the quarter and 35.7% for the nine months, and net income for nine months substantially exceeded the prior year. Backlog stabilised at $137.1 million. Unfunded backlog had fallen to approximately $14.5 million.

Most notably, the FY2026 guidance was revised upward from the September 2025 position. Management now stated it expected higher revenues than FY2025 and that net income for FY2026 would exceed FY2025 - replacing the earlier "net income per share will decline due to higher costs in the product mix."

"With a strong quarter behind us, we continue accelerating into the remainder of the year with confidence" in delivering "excellent full-year results." - David O'Neil, Q3 FY2026 release, May 12, 2026

Management also disclosed the $152.5 million bid pipeline and the explicit statement that FY2026 order intake would finish below FY2025 - again, volunteering the negative.

Promise versus outcome

Assessment

This management does what it says, and it errs conservative. That is an unusual verdict and it rests on one hard fact: in September 2025 management told the market that FY2026 earnings per share would decline, at a moment when nobody was asking, and it then beat that guidance so decisively that it had to revise it upward eight months later. Companies that manage expectations upward do not do that. Companies that overpromise do not do that. The cost-savings commitment made in February 2025 was delivered and traceable in margin across four consecutive periods. The unfunded-backlog conversion happened as described.

The one genuine criticism is a narrative one, not a disclosure one. In the Q2 FY2026 release, the CEO's quoted commentary said "encouraging trends" while the same release disclosed that six-month order intake had fallen roughly two-thirds. The number was there in black and white - Espey did not hide it - but the CEO chose not to speak to it. If Espey held earnings calls, that is the question an analyst would have asked, and the absence of anyone to ask it is the structural weakness in assessing this management at all.

Verdict: credible and conservative on everything that can be measured, with a mild tendency to let the numbers deliver bad news while the quotes deliver good news. Given a two-year record of beaten guidance, that is a benign form of the failing.

10. Shareholder friendliness index

Dividends. Espey has raised its regular dividend in every one of the last three fiscal years and added a special on top. FY2024 regular dividends totalled $0.675 per share ($0.15 in September 2023, $0.15 in December 2023, $0.175 in March 2024, $0.20 in June 2024). FY2025 totalled $1.00 per share ($0.25 in each of the four quarters). FY2026 totalled $1.75 per share - four quarterly payments of $0.25 plus a $0.75 special cash dividend declared September 8, 2025 with a September 19, 2025 ex-date (dividend history; Espey press releases). The most recent declaration was the regular $0.25 announced June 4, 2026. That is a 159% increase in the regular dividend over three fiscal years, plus a special. The special is the tell: it followed the year in which Espey booked $86.4 million of orders and record backlog, and it is how a company with no debt and a large cash and investment balance returns a windfall without committing to a permanently higher run-rate.

Buybacks and dilution. Here the record is the opposite. A board repurchase authorisation exists, but the remaining authorisation has stood at exactly $783,460 in every filing reviewed from FY2024 through the Q3 FY2026 10-Q - unchanged, meaning nothing has been bought under it in that window. The FY2025 10-K states there were no open-market purchases in Q4 FY2025 and no ESOP share repurchases in FY2025, and the Q3 FY2025 and FY2024 quarterly filings likewise record no ESOP repurchases (FY2025 10-K). MoatMap's database records zero buybacks in the trailing ~90 days to August 21, 2026, consistent with the filing record but covering only that window. Authorised: $783,460 remaining. Actually repurchased over three years: nothing I could locate in any filing. Meanwhile shares outstanding went 2,702,633 at June 30, 2023 → 2,733,958 at June 30, 2024 → 2,896,368 at June 30, 2025 → approximately 2.96 million currently - roughly +9.5% over three years, entirely from stock option exercises (31,325 options were exercised in FY2024 alone, and Section 11 documents heavy exercise activity through FY2026). The share count is growing, not shrinking, and the growth is accelerating as an appreciating share price pulls old low-strike options into the money.

Verdict: Returns Capital - via dividends only. The regular dividend has risen 159% in three years and was topped with a special, but shareholders are simultaneously being diluted roughly 3% a year by option exercises that the untouched buyback authorisation is not offsetting.

11. Insider activities

Source and method. The MoatMap database block covers the trailing 12 months from a US venue (SEC Form 4), and is the spine of this section. Because the US is an open venue, I cross-checked the most recent weeks and the full 12-month shape against the SEC Form 4 record as aggregated at secform4.com for CIK 33533 and against individual Form 4 filings. MoatMap data is current as of 2026-08-21 23:00 UTC; the cross-check found no filings newer than the August 4, 2026 transaction already in the block.

One labelling discrepancy to disclose up front. The August 4, 2026 and June 22, 2026 Michael Wool transactions are recorded by MoatMap as "Other" at reference prices of $8.98 and $11.25, and by the Form 4 aggregator as "Gift" at the same prices. Those two labels are not reconcilable from the summary data - a Form 4 gift (code G) carries no price, while $8.98 and $11.25 look like legacy option strike prices (code M). I have not been able to resolve which it is from the aggregated records. What is unambiguous either way: neither was an open-market purchase.

Recent transactions (12 months to August 2026)

Sources: MoatMap multiverse insider database (US venue, SEC Form 4), cross-checked against secform4.com CIK 33533 and individual Form 4 filings including Helmetag, March 2026.

MoatMap's aggregate for the 12-month window: 93 transactions, 5 distinct insiders, 0 buys / 84 sells / 9 other. The high transaction count is an artefact of how these sells were executed - Corr's June 23, 2026 disposal alone was broken into roughly 30 individual fills between $65.08 and $66.40, and Helmetag's June 24 sale into 24 fills between $63.13 and $63.98. Five distinct people, not ninety-three decisions.

Buys - read the signal

There were none. Across the full 12 months, not one insider made an open-market purchase of Espey stock. Zero, from five active insiders including the CEO and the Chairman.

This deserves to be stated flatly rather than softened. Open-market buying is the highest-signal item in this section precisely because insiders have many reasons to sell and only one reason to buy. Espey's insiders produced none of it during a period in which the company reported record backlog, converted $18.5 million of unfunded backlog to funded, brought a Navy-funded facility into production, and revised full-year guidance upward. If management believed the shares were mispriced against that operating record, the buying window was wide open and nobody used it.

Sells - working out the why

The pattern is unusually legible once you line up the strike prices against the sale prices.

Every large disposal in this window is an option exercise-and-sell, not a position liquidation. Corr exercised at $18.05 and sold at $65-$66. Corr exercised at $20.50 and sold at $67.06. Pickering exercised at $24.46 and sold at $63.00. O'Neil exercised at $21.75-$26.25 and sold at $50.00. These are options struck years ago, at prices between roughly $13.61 and $27.21, being converted at market prices of $43 to $68. The stock has roughly tripled off the strike levels. The economic content of these transactions is the realisation of long-dated compensation, not a change of view.

The specific reasons, insider by insider:

• David A. O'Neil (President and CEO). The heaviest participant. Three pure option exercises in August-September 2025 at strikes of $13.61, $18.05-$20.50 and $16.54, totalling 16,250 shares acquired; two open-market sales in December 2025 at $43.75 and $46.27 totalling 4,000 shares; a large 11,656-share exercise-and-sell in February 2026 at $50.00; and a 2,500-share open-market sale in June 2026 at $67.71. Reason not disclosed in any of these filings. The most plausible read from the record itself - and I am labelling it a read, not a disclosure - is a staged monetisation of deep-in-the-money options that had appreciated dramatically, executed in tranches as the price rose. What the record does not show is a 10b5-1 plan reference on the transactions examined, which would have provided a clean, pre-scheduled explanation.

• Carl Helmetag (Chairman of the Board). Exercises at $26.25 (March 2026) and $21.75 (May 2026); open-market sales of 1,000 shares in December 2025 at $44.99 and 1,000 shares in June 2026 at ~$63.66. Crucially, he retains a substantial position: 23,169 shares direct plus 1,504 in an IRA after the June 2026 sale, and 16,542 options remaining after the March exercise. A director selling 1,000 shares while holding 23,169 is trimming, not exiting. The March 2026 Form 4 explicitly shows the 10b5-1 box unchecked, meaning that particular exercise was not made under a pre-arranged plan - a small negative, since a plan would have removed any question of timing discretion.

• Paul J. Corr (Director). The most active seller by transaction count - three exercise-and-sell cycles in February, May and June 2026, each of roughly 2,550-4,642 shares, at strikes of $18.05-$27.21 against sale prices of $57.90-$67.06. He filed a Form 144 on February 18, 2026, the standard notice for a proposed sale of restricted or control securities, which is procedurally routine. Contextual note: Corr stepped down as Audit Committee Chair on December 5, 2025, succeeded by Nancy Patzwahl, while continuing as a board member and Audit Committee member (Espey press release, December 8, 2025). A committee-chair handover often accompanies a director's own planning horizon, but the filings state no reason and I will not manufacture one. Reason not disclosed.

• Jennifer Michele Pickering (CHRO and Corporate Secretary). A single 2,500-share exercise-and-sell on June 22, 2026, at a $24.46 strike against a $63.00 sale. This is the textbook shape of an executive converting vested compensation - one transaction, full exercise-and-sell, no residual open-market selling. Reason not disclosed, but the shape is unremarkable.

• Michael W. Wool (Director). The odd one out. His three transactions - 1,842 shares exercised at $27.21 in February 2026, 2,150 at $20.50 in May 2026, and the two disputed-label transactions in June and August 2026 at $11.25 and $8.98 references - include exercises with no accompanying sale recorded. An insider who exercises and holds is doing the opposite of an insider who exercises and sells: he is paying cash to increase his equity exposure. That is a mildly constructive data point buried inside an otherwise one-directional table, and it is the reason the two disputed-label items matter enough to flag rather than ignore.

The clustering, and what it actually reflects

Four of the five insiders transacted inside a three-day window, June 22-24, 2026 - O'Neil, Pickering, Corr and Helmetag, plus Wool's June 22 item. Cluster selling of that tightness would normally be a serious warning. Here the more economical explanation is a common trading window opening after the fiscal-Q4 blackout arrangements and ahead of the June 30 year-end, with several people using it for compensation realisation at what was then an all-time-high share price of $63-$68. The company reported nothing adverse in the following weeks, and no 8-K, guidance change or contract loss has surfaced since.

That said, "everyone sold at the high together" and "everyone used the same window" are observationally identical from outside, and Espey - having no earnings call - offers no forum in which anyone could ask.

Net assessment

Espey's insiders were decisively net sellers over the last 12 months: 84 sale fills, 9 non-directional transactions, and zero open-market purchases from five insiders. Activity is broad-based rather than concentrated - the CEO, the Chairman, two other directors and the CHRO all participated - which removes the "one person's personal circumstances" explanation that often defuses a selling record.

Three things temper the reading. First, essentially all of it is option exercise-and-sell on strikes struck at a third to a half of current prices, which is compensation realisation rather than a portfolio decision. Second, Helmetag retains 23,169 shares plus 16,542 options after selling, and Wool exercised without selling - neither is the behaviour of someone exiting. Third, the selling coincides with the shares reaching all-time highs after roughly tripling, which is when option holders mechanically monetise regardless of view.

Two things sharpen it. First, the complete absence of any open-market buy over a full year, during a period of record backlog and upgraded guidance, is a real negative - it is the cheapest possible way for management to express confidence and nobody spent a dollar on it. Second, the selling accelerated into the June 2026 highs across four insiders simultaneously, and at least one Form 4 examined shows no 10b5-1 plan, so timing discretion sat with the insiders rather than with a pre-committed schedule.

Plain-language read: mild concern. Not a red flag - the mechanics are explicable and two insiders demonstrably retain or increased exposure. But a year of exercise-and-sell with zero conviction buying, clustered at all-time highs and without the cover of pre-arranged plans, is the weaker end of neutral rather than the stronger end.

12. Scenarios

Bull case

The $152.5 million bid pipeline converts. Espey's FY2026 order drought turns out to have been exactly what management implied it was - the trough between two award cycles, with FY2025's $86.4 million intake having pulled forward the Columbia and Virginia work that would otherwise have been booked across two years. The Navy's FY2027 budget request, with its 46% increase in shipbuilding and $6.2 billion earmarked for submarine industrial-base capacity, starts landing on suppliers in the second half of calendar 2026 and through 2027. Espey, having just been handed $10.8 million of government money to build and equip a magnetics facility, is on the short list of firms the Navy is explicitly trying to grow.

The Magnetics Center of Excellence gets fully staffed and becomes what it was funded to be: not just more floor space but a capacity constraint removed on the one product family where Espey's competitive position is close to unassailable. Higher-value naval transformer work displaces lower-value build-to-print in the mix, and the margin improvement visible through FY2026 turns out to be structural rather than a favourable-mix accident. Espey's own pitch - that it holds large-business quality disciplines inside a small-business cost structure - becomes the reason primes bring it new designs rather than only repeat production.

The design-in flywheel then does the compounding. Each new part Espey qualifies on a Virginia or Columbia hull, or on a new surface combatant class, becomes a two-decade annuity that no competitor can economically challenge, because challenging it means re-running MIL-S-901D shock qualification on a part worth a few hundred thousand dollars a year. Customer concentration keeps drifting down as the pipeline brings in programs beyond the two that dominate today. The dividend, already up 159% in three years and topped with a special, keeps climbing, and the board finally uses the untouched buyback authorisation to stop the option-driven share creep.

By 2029 Espey is a company with a larger installed base of qualified part numbers than it has today, spread over more programs, produced in a facility that the customer paid to build.

Base case

Espey delivers roughly what management has already told the market. FY2026 finishes with revenue and net income above FY2025 - the guidance given in May 2026 - and the September 2026 year-end release confirms full-year order intake well below FY2025's $86.4 million, exactly as management pre-announced. Backlog settles somewhere near or modestly below the $137 million reported at March 31, 2026, because shipments continue to run ahead of a normalised order rate.

FY2027 is the heavy conversion year - 38% of the March 2026 backlog is scheduled to ship in it - so revenue holds up and the factory stays busy regardless of what happens to intake in the meantime. Order flow recovers to something between the $52 million of FY2024 and the $86 million of FY2025 as the Navy's expanded shipbuilding budget filters down to suppliers, without any single award matching the $29.5 million Columbia panel order. The bid pipeline converts partially, at normal hit rates, on a mix of repeat and new programs.

Operationally, life stays lumpy. A quarter or two each year gets disrupted by component lead times approaching a year, by a continuing resolution, or by a prime moving a delivery date, and management explains it in the same measured language it has used for six consecutive periods: a timing shift, not lost business. The Magnetics Center of Excellence ramps but is staffing-limited rather than floor-space-limited, so capacity converts to output more slowly than the square footage would suggest. Customer concentration stays in the 60-75% range across four to six accounts because the customer universe is genuinely that small.

The regular dividend keeps rising, specials appear in windfall years, the buyback authorisation stays untouched, and the share count keeps drifting up a few percent a year as legacy options are exercised. Espey remains what it has been for decades: a small, cash-rich, single-facility company that is structurally important to programs far larger than itself.

Bear case

The order drought is not a trough - it is the new baseline. The $29.5 million and $19.8 million awards turn out to have been the once-a-decade events they look like, having captured the full multi-year buy for the parts Espey owns on Virginia and Columbia. The $152.5 million bid pipeline converts at a poor rate because most of it is new-design work where Espey is not the incumbent and is bidding against primes' in-house power groups and better-capitalised merchant vendors. FY2028 arrives with 29% of the old backlog left to ship and nothing meaningful behind it, and a company that spent two years looking fully booked suddenly looks empty.

Simultaneously, the fixed-price engineering exposure bites. The $13 million of engineering programs identified in the FY2025 10-K includes at least one development contract - plausibly in the directed energy or energy storage line, where Espey has the least track record - that runs into a technical problem. On a fixed-price contract with no portfolio to absorb it, a single bad program dominates a reporting year, and a company with no analyst coverage and no earnings call discovers its shareholder base has no patience for a surprise it was given no forum to anticipate.

The structural erosion runs underneath both. Wide-bandgap semiconductors keep shrinking the magnetics content per system. Merchant vendors with vastly larger R&D budgets push MIL-COTS modules further into applications that once required a custom Espey design, and the build-to-print business - the portable, drawing-package half of the catalogue - migrates to lower-cost contract manufacturers. Espey's genuine moat contracts to a narrowing set of submarine part numbers.

The final compression comes from the things a 152-person company cannot buy its way out of. Skilled winders and test technicians do not materialise in Saratoga Springs, so the Navy-funded facility stands partly idle. The IBEW agreement expires June 30, 2028 into a tight labour market. DFARS cybersecurity requirements keep ratcheting against a fixed cost base that cannot spread them. A top customer - and five customers are 73% of a quarter's sales - decides to make the part in-house or loses the program. And through all of it, the shareholder record shows that when the shares were at all-time highs and the operating news was at its best, not one insider bought a single share in the open market.