Eisenhower’s line gets quoted often enough that it can start to feel like a slogan. It isn’t. Move a brigade, sustain a carrier strike group, or keep a fifth-generation fighter mission-capable, and the truth becomes obvious very quickly: logistics is not the supporting act. It is the show. And inside logistics sits an unglamorous discipline that quietly determines whether everything else works — inventory management. Knowing what you have, where it is, whether it is real, and whether it is the exact item you think it is — the one it should be.

These questions are where modern military logistics has a problem hiding in plain sight. The DoD has spent years of cybersecurity investment that more recently coalesced around a Zero Trust architecture, in which no user, device, or packet is assumed legitimate on the network without continuous verification.[1] Meanwhile, the physical items that move through DoD supply chains — weapons, night-vision devices, avionics modules, and the very circuit boards cyber may be running on — are still verified by the digital equivalent of a name tag. A sticker. A barcode. A laser etch. A piece of paperwork riding along in the crate.

We trust the proxy, not the item. And the proxy can be peeled off, photocopied, swapped, faked, or simply worn away in the field. It is time to retire the proxy.

The Proxy Problem

Every method the military currently uses to identify a physical item is, at bottom, an additive marker stuck onto the item: serial numbers stamped or engraved, RFID tags glued or embedded, QR codes printed on labels, Item Unique Identification (IUID) data matrix marks lasered onto a surface, manufacturer’s nameplates riveted on. These methods are not worthless — they have served logistics for generations. But they share a single, structural weakness. The marker is not the item. It is a representation of the item, and that representation can be separated from the thing it represents.

A tag falls off a pallet in a dusty motor pool and the item underneath becomes effectively anonymous. A QR code is reprinted by a counterfeiter in a back room overseas and a fake circuit board ships through three intermediaries with a perfect paper trail. A serial number on an engraved plate is ground off and re-stamped. A unit-level inventory check at 0200 hrs sees an etched mark on a weapon and trusts that the etching tells the truth about the weapon underneath.

The proxy assumes integrity it cannot actually deliver. And the consequences have been well documented. In April 2024, the US Department of Justice secured a six-and-a-half-year prison sentence for the leader of a scheme that imported roughly $1B worth of counterfeit Cisco networking equipment from China and Hong Kong and sold it into US government supply chains. The fraudulent gear — old, lower-grade devices physically modified and re-labeled to look like new, higher-end Cisco products — ended up in classified information systems and on platforms supporting USAF F-15s, B-52, and other military aircraft. Customs and Border Protection had intercepted around 180 shipments over the preceding eight years, and the trafficker had responded to multiple Cisco cease-and-desist letters with forged documents. None of this required a sophisticated cyberattack. It required only the manipulation of a proxy: a label, a serial sticker, a paperwork entry.[2]

The Cisco case is the adversarial failure mode. There is also a quieter, non-adversarial version that the same proxy model produces independently. In May 2023, the Government Accountability Office (GAO) reported that more than one million F-35 spare parts — worth at least $85M — had gone missing across the program’s global supply chain since 2018. The GAO found that the DoD lacked its own system to account for these parts, that the prime contractors used their own incompatible inventory systems, and that DoD had reviewed the circumstances of only about 2% of the identified losses. The F-35 program is projected to cost $1.7 trillion over its life. Proxy-based identification allows millions of supply chain components to simply vanish without any active tampering or adversary intervention. Add an attacker, and the numbers get worse.[3]

Cybersecurity has moved away from this kind of perimeter-and-credential model precisely because adversaries learned to steal credentials. Zero Trust says: never assume the credential is the user. Continuously verify. Yet on the physical side, we still treat the sticker as the part.

How We Solved This Problem for Humans

There is a useful parallel sitting in our own pockets. For most of modern history, identifying a person meant identifying a proxy attached to that person. A name written in a ledger. A signed letter of introduction. A passport. A driver’s license. A military ID card. An employee badge. The state-of-the-art was to make the proxy harder to forge — watermarks, holograms, microtext, embedded chips.

It never quite worked. Passports were stolen and altered. ID cards were cloned. Credentials were borrowed by people who looked enough like the old photo. Every increase in proxy security was met by an increase in fraud sophistication. This “arms race” never ended because the underlying model was wrong: we were trying to authenticate the card, not the person.

Then biometrics changed the question. Instead of asking whether the card is real, ask whether the person is who they claim to be — by reading something inherent, unique, and persistent to that person that cannot be detached from them. A fingerprint. An iris pattern. A facial geometry. A voiceprint. These features are not added to a human. They are the human. They were there before the credential was issued and they will be there after the credential expires. Reading them does not require trusting any document or marker. It requires only the ability to observe the person closely enough to extract a stable, distinguishing signature.

That shift — from authenticating the proxy to authenticating the entity itself — is now standard for high-stakes person identification. Customs and Border Protection runs facial comparison at airports. The FBI maintains biometric databases at scale. Technicians enroll fingerprints and iris scans. Phones unlock by face. None of this asks you to trust your ID card. It asks the system to read you.

Physical items, especially critical and sensitive ones, deserve the same upgrade.

Items Have Fingerprints Too

Here is the part many logisticians do not yet appreciate: every manufactured item is already unique. Manufacturing tolerances guarantee it. The allowable variation in dimensions, surface texture, material composition, machine vibration during cutting, tool wear, ambient temperature, operator handling — all of it conspires to ensure that two parts coming off the same line, from the same machine, in the same batch, while identical from a practical operational sense, are in-fact physically different at fine scale. Not different enough that they fail their drawing or quality check. Different enough that, examined closely, they are distinguishable.

It is the snowflake principle, applied to industry. Two snowflakes that form an inch apart in the same cloud are individually unique because the precise path of formation differs in microscopic ways. Two stainless-steel fasteners coming off the same screw machine are individually unique for the same kind of reason — each one acquires a microsurface signature that nothing else in the world has.

Until recently, that uniqueness was a curiosity. We had no practical way to read it at scale. The optics, the compute, and the machine-vision algorithms simply weren’t there. They are now. A standard industrial camera, or in many cases an off-the-shelf smartphone, paired with modern computer-vision models, can extract from a small region of a part’s surface, a signature stable enough and distinctive enough to identify that specific item against a population of millions of nominally identical siblings with extraordinary accuracy. The item enrolls itself once. From then on, any subsequent image of it can be matched back to that enrollment without consulting any tag, label, or paperwork.

This is the physical-world equivalent of biometric identification. The item is its own ID. No proxy required. No marker to fall off. No serial number to scratch off. No QR code to reprint in a counterfeiting workshop. The identifier is the part.

What This Changes for the Military

Adopting item-level, inherent identification — the biometric model, applied to things — reshapes several pain points that the proxy model has never solved well.

Counterfeits stop being a labeling problem

A counterfeit Cisco board enters the DoD network because its label matches the real one. A counterfeit aerospace engine part is accepted because its packaging is convincing. Under an inherent-identification regime, the question shifts. The system does not ask whether the markings match. It asks whether this physical board, with its specific surface signature, is the one that was enrolled at the trusted supplier. If it isn’t, it doesn’t matter how good the label looks. The item itself fails authentication.

Sensitive item accountability becomes faster and harder to fool

Monthly weapons and night-vision device inventories at the unit level today involve supply technicians reading engraved serial numbers against a hand-written or scanned roster. The process is slow, error-prone, and depends entirely on the integrity of the etched mark and many times, the handwriting of the Warfighter who quickly filled out the form. Replacing the read of the etching with a quick visual scan of the item itself — capturing the item’s natural surface signature — collapses the time required and removes the error and failure modes from “fat fingering” to fraud.

Lifecycle traceability extends from cradle to grave

A part enrolled at the manufacturer can be re-identified at acceptance, at the depot, at the forward operating base, at maintenance, at overhaul, and at disposal — without ever depending on a tag surviving that journey. Tags do not survive that journey. Items do. This is exactly the kind of persistent, non-transferable identity that asset serialization and tracing has wanted for decades and never been able to deliver through additive markers, no matter how sophisticated.

Human error in inventory drops

A significant share of the inventory discrepancies the GAO has documented across DoD components trace back to manual processes — misreads, miskeyed serial numbers, transposed digits, mislabeled crates. When the system reads the item directly, those failure modes disappear. The technician points a camera at the part. The part is uniquely identified. There is no number to read or retype.

Zero Trust finally extends to the physical layer

DoD has invested heavily in Zero Trust for its cyber networks. The principle is straightforward: continuously verify, never assume. Applying that same principle to physical assets has been blocked by the fact that we had no way to verify a physical item without consulting a proxy credential attached to it. Inherent identification removes that block. Every item, every time it is handled, can be verified against its enrolled signature. The physical supply chain finally operates on the same zero trust model as the network.

This Is Not a New Mandate — It Is a Better Way to Meet Existing Ones

DoDI 4140.01 already requires effective supply chain materiel management.[4] DoDI 4140.67 already mandates counterfeit prevention.[5] IUID policy already requires that DoD assets above certain thresholds carry persistent, unique identifiers.[6] The IUID program has been in place for two decades. Its central instrument — the data matrix mark — is, again, an additive proxy. It is a more sophisticated proxy than a stamped serial number, but it remains a marker affixed to the item and not the item itself.

Staff Sgt. Aaron Ray, 96th Logistics Readiness Squadron, scans an electronic shelving label April 11 at Eglin Air Force Base, Fla. The label can display whatever the 96th LRS’s warehouse personnel needs to display. The labels saved the warehouse staff approximately 5,000 hours by streamlining their storage procedures. (U.S. Air Force photo/Samuel King Jr.)

An inherent-identification approach is not in tension with IUID. It is the natural next step of IUID. The unique identifier still exists. It is still persistent. It is still readable by a handheld device. What changes is where the identifier lives, or more accurately, what it is: not on a sticker that can come off, but in the surface of the item itself. The policy intent — persistent, unique, machine-readable identification of every relevant asset — is finally achievable without the operational fragility that has always limited proxy-based methods in the field.

The same logic applies to readiness reporting under DoDI 7730.66.[7] Readiness assessments are only as honest as the underlying property accountability. If the items being counted cannot be reliably distinguished from their counterfeit or substituted twins, or even those acquired in error, readiness numbers reflect what the system was told, not what is actually in the supply chain. Inherent identification raises the floor on the integrity of that data.

The Implementation Reality

Skeptics in any logistics discussion will, reasonably, ask the operational questions. Does it work in austere environments? Does it require specialized equipment? Does it survive contact with how Warfighters actually work? These are the questions that have killed many promising logistics technologies over the years.

The answer is that machine-vision identification of items has matured to the point where commercial deployments in aerospace, healthcare, luxury goods, and high-value industrial supply chains are running at high accuracy using standard industrial cameras and, in many cases, ordinary smartphones. There is no requirement for clean rooms, special lighting rigs, or operator certifications measured in weeks. A technician with a phone can enroll and verify items in the field with training measured in minutes.

There are real questions worth working through. How do enrollment databases scale across coalition partners? How do classification considerations interact with cloud-based or distributed identification infrastructures? How does the policy framework around chain-of-custody evidence evolve when the evidence is a surface signature instead of a serial number? These are good questions. They are not blockers. They are the kind of integration questions that DoD has answered before for every previous shift in identification technology, including the move to IUID itself.

The Strategic Argument

Zoom out from the warehouse for a moment. The competitive landscape for the next two decades will be defined in significant part by supply chain resilience. Adversaries already understand this. The pattern of counterfeit component introductions, the targeted disruption of critical material flows, the exploitation of brittle authentication processes — these are not random crimes. They are part of a coherent effort to degrade US and allied capability at lower cost and lower escalation risk than a kinetic confrontation.

A defense supply chain that authenticates its physical assets the way it authenticates its network traffic is a harder target. Counterfeits cannot survive an authentication step that does not depend on the marker the counterfeiter controls. Insider substitution becomes detectable because the substituted item fails to match its enrolled signature. Lost or stolen sensitive items can be re-identified if and when they reappear, regardless of what has been done to their serial numbers.

There is also an allied-interoperability case. Coalition operations routinely run into the problem that two nations operating the same equipment cannot easily reconcile their inventories because their proxy systems differ — different label formats, different RFID standards, different paperwork. An identification approach that reads the item itself bypasses that incompatibility. The item is the same item regardless of whose label it is wearing today.

The Path Forward

The shift from proxy-based to item-based identification will not happen overnight. Neither did the shift from passports to biometrics. Civil aviation took roughly two decades to move from “show us your ID” to “stand in front of the camera.” It happened because the alternative — trusting documents in an era of sophisticated forgery — stopped being defensible.

The same logic now applies to physical items in the defense supply chain. Trusting tags, labels, etchings, and printed codes in an era of sophisticated counterfeiting and well-resourced adversaries is no longer defensible. The technology to do better exists and is in commercial use. The policy framework largely permits it and in many cases already demands its objectives. The training and infrastructure burden is low. What is required is the decision to treat physical items the way we have learned to treat people: as unique entities to be identified on their own terms, not as bearers of credentials we hope no one has tampered with.

Eisenhower’s line about logistics was not a celebration. It was a warning. Wars are won and lost on logistics because the side whose logistics breaks first, loses. The proxy is where ours breaks. It is past time to fix it.

About the author

Roei Ganzarski is a technology executive with three decades of experience in aerospace, defense-adjacent industries, and supply chain technology. He spent more than a decade with The Boeing Company, where he served as Chief Customer Officer for Boeing’s Flight Services division. He later led BoldIQ, a real-time scheduling optimization software company, and served as CEO of magniX and Executive Chairman of Eviation Aircraft, where he helped pioneer the electric aviation industry. He has testified before the US Congress on the future of aviation and was named to Business Insider’s “100 People Transforming Business.” He currently serves as CEO of Alitheon, an optical-AI company working on item-level serialization of physical goods — the technology category discussed in this article. He holds an MBA from the University of Washington, is a graduate of Wharton’s Advanced Management Program, and is completing his doctorate in innovation and critical thinking at SDA Bocconi.

Work Cited

[1] U.S. Department of Defense, DoD Zero Trust Strategy, DoD CIO Zero Trust Portfolio Management Office, October 21, 2022, https://dodcio.defense.gov/Portals/0/Documents/Library/DoD-ZTStrategy.pdf.

[2] U.S. Department of Justice, Office of Public Affairs, “Leader of Massive Scheme to Traffic in Fraudulent and Counterfeit Cisco Networking Equipment Sentenced to Prison,” press release, May 1, 2024, https://www.justice.gov/opa/pr/leader-massive-scheme-traffic-fraudulent-and-counterfeit-cisco-networking-equipment.

[3] U.S. Government Accountability Office, F-35 Program: DOD Needs Better Accountability for Global Spare Parts and Reporting of Losses Worth Millions, GAO-23-106098, May 23, 2023, https://www.gao.gov/products/gao-23-106098.

[4] Department of Defense Instruction 4140.01, DoD Supply Chain Materiel Management Policy, https://www.esd.whs.mil/Portals/54/Documents/DD/issuances/dodi/414001p.pdf.

[5] Department of Defense Instruction 4140.67, DoD Counterfeit Prevention Policy, April 26, 2013, incorporating change 3, effective February 2, 2024, https://www.esd.whs.mil/Portals/54/Documents/DD/issuances/dodi/414067p.pdf.

[6] Department of Defense Instruction 8320.04, Item Unique Identification (IUID) Standards for Tangible Personal Property. See also Defense Acquisition University, “Item Unique Identification (IUID),” https://www.dau.edu/acquipedia-article/item-unique-identification-iuid.

[7] Department of Defense Instruction 7730.66, Readiness Reporting Guidance for the Defense Readiness Reporting System.