Industries / Transport and vehicles / Aerospace and defense / Defense contracting
Logistics and supply
Logistics and supply is the work of buying, storing, tracking and moving the spare parts, food, fuel and equipment that armed forces use, done by government supply agencies and the contractors that work for them.
5
Milestones achieved, from 1958 to 2012.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1958
NATO nations share spare parts
In 1958 the North Atlantic Council approved a NATO Maintenance Supply Services Agency, which NATO says changed its name five years later to the NATO Maintenance and Supply Agency (NAMSA). Its successor, formed in Luxembourg on 1 July 2012 from three support agencies and named the NATO Support and Procurement Agency (NSPA) in April 2015, describes its work as "multinational support solutions" and runs the NATO Logistics Stock Exchange, where customers arrange the exchange of excess spare parts.
Credit
North Atlantic Council; NAMSA
1962
One agency for common supplies
From the mid-1950s each group of common items was managed for all US services by one "single manager" agency: the Army ran food and clothing, the Navy medical supplies, petroleum and industrial parts, and the Air Force electronic items. Each manager worked under its own service's procedures, so customers needed as many sets of procedures as there were managers. Secretary of Defense Robert McNamara ordered them merged: the Defense Supply Agency was established on 1 October 1961 and began operations on 1 January 1962, taking in eight single-manager agencies. It took over worldwide bulk fuel in 1972 and troop food in 1973, and was renamed the Defense Logistics Agency (DLA) on 1 January 1977.
Credit
Robert McNamara; Defense Supply Agency
2003
Gulf War lessons repeated
On 18 December 2003 the US General Accounting Office (GAO, now the Government Accountability Office) reported early findings on supply in the Iraq war: backlogs of hundreds of pallets and containers, a US$1.2 billion gap between materiel shipped to Army units and materiel they acknowledged receiving, duplicate orders, and containers that had not been opened or inventoried. Radio frequency identification (RFID) tags, which broadcast what a container holds, were required where feasible but were not used consistently, and staff had not been trained on them. GAO noted that its 1992 report had found the same loss of visibility over supplies in the 1991 Gulf War.
LessonTracking tools change nothing unless every shipper uses them and every unit can read them.
Credit
GAO
2012
Fake chips bought online
GAO investigators set up a fictitious company, joined two internet platforms that sell military-grade electronic parts, and bought 16 parts. An independent lab judged all 12 parts ordered under real part numbers "suspect counterfeit", and four vendors shipped parts for part numbers GAO had invented. Of the 396 vendors that answered, 334 were in China. Section 818 of US Public Law 112-81 and the defense purchasing clause that applies it (DFARS 252.246-7007) now require larger contractors to run a system for detecting and avoiding counterfeit electronic parts; without one, payments can be withheld and the cost of counterfeit parts and rework disallowed.
LessonBuying obsolete parts from unknown brokers invites fakes, so the law moved the duty to check onto contractors.
Credit
GAO; US Congress
2012
Stock that never moved
The UK National Audit Office (NAO) found that the Ministry of Defence's non-explosive stock rose 13%, from £17.2 billion in March 2009 to £19.5 billion in December 2011. Of £4 billion spent on raw materials and consumables from April 2009 to March 2011, £1.5 billion (38%) was not used; £4.2 billion of stock had not moved for at least two years; and storing and managing it cost at least £277 million in 2010-11, by the NAO's estimates. Amyas Morse, head of the NAO, said that "management and accountability structures currently fail to provide the incentives for cost-effective inventory management".
LessonWithout targets that reward lean stock, depots over-order.
Credit
NAO
Watch for
Robot-formed panel ends a wait
A US Air Force C-17 transport damaged in a storm on 4 March 2025 needed a new left nose panel, and Boeing, its maker, found no vendor willing to build tooling for a single part. The Air Force Rapid Sustainment Office and the University of Dayton Research Institute made one by incremental sheet forming, in which a robot presses a flat metal sheet into shape in many small steps without a die. The aircraft flew again on 30 July 2026. The Air Force says this is the first such part flown on one of its aircraft, that it cut a repair of years to months, and that parts for the KC-135 and F-15 are in work.
Achieved whenthe process supplies parts for other aircraft types in routine use, or an outside body such as GAO reports its cost and time against normal sourcing.
Claimed by
Air Force Life Cycle Management Center, August 2026
A field-printed part in hours
SPEE3D, an Australian maker of metal 3D printers, says soldiers of the Tennessee Army National Guard, helped by University of Tennessee engineers, designed, printed, heat-treated and machined a replacement door handle for a disabled combat support vehicle during a field exercise, using its cold spray process, a form of metal 3D printing, and that the part was made and fitted in under 10 hours. It compares this with a supply process that can take six to ten weeks. The university, the Guard and the Army Research Laboratory received an award for the project at the Military Additive Manufacturing conference in 2026.
Achieved whenan army or an independent lab publishes how field-printed parts were approved and how they performed over many repairs, not one exercise.
Claimed by
SPEE3D, March 2026
Open
01
Offers for invented part numbers
A repeatable public count of how often online electronic-parts marketplaces and broker sites show stock for part numbers that were never made. Write 200 invented numbers that follow real makers' numbering patterns, plus 200 real numbers for parts discontinued more than 15 years ago as a comparison. Search each on public broker platforms with a scripted browser and record which return "in stock" listings, with the quantities and date codes claimed. Observe listings only: no quote requests, no purchases and no invented company. Publish counts by platform, not seller names, and follow each site's terms of use.
The briefCredit · unclaimed02
A ledger of repeated findings
A public table of the supply and logistics findings in published audit reports on deployed operations since 1991, from GAO, the UK National Audit Office, the Australian National Audit Office and Canada's Auditor General. Code each finding by problem type (lost track of cargo, unlabeled containers, duplicate orders, untrained users of tracking tools, systems that cannot exchange data) and record the report, the operation, whether the recommendation was marked as carried out, and whether a later report found the same problem. Keep to management findings already made public; leave out anything on routes, stock levels or readiness.
The briefCredit · unclaimed
Open 01 · the brief
Offers for invented part numbers
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A repeatable public count of how often online electronic-parts marketplaces and broker sites show stock for part numbers that were never made. Write 200 invented numbers that follow real makers' numbering patterns, plus 200 real numbers for parts discontinued more than 15 years ago as a comparison. Search each on public broker platforms with a scripted browser and record which return "in stock" listings, with the quantities and date codes claimed. Observe listings only: no quote requests, no purchases and no invented company. Publish counts by platform, not seller names, and follow each site's terms of use.
Why it is still open
GAO's 2012 test is the last public one found. Asked to quote on invented part numbers, four vendors sold GAO parts anyway. A search found no later or larger repeat, and GAO's method of posing as a buyer is not one a private tester should copy, so a listing-only method is still to be built.
Built with what exists
Python and Playwright; a browser agent such as Browser Use for sites that need form searches (see the Note "Deleting 80% to start over"); makers' public part-number guides to build plausible fake numbers; SQLite; GitHub to publish the code and counts.
First step
Write 20 invented numbers for one common part family, such as a voltage regulator series, confirm with the maker's own parametric search that none exist, and search them on three platforms.
Done looks like
400 part numbers searched on at least 10 platforms, repeated monthly for three months, with the share of invented numbers shown as in stock on each platform, and a second person rechecking 10% of the hits by hand.
Open 02 · the brief
A ledger of repeated findings
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of the supply and logistics findings in published audit reports on deployed operations since 1991, from GAO, the UK National Audit Office, the Australian National Audit Office and Canada's Auditor General. Code each finding by problem type (lost track of cargo, unlabeled containers, duplicate orders, untrained users of tracking tools, systems that cannot exchange data) and record the report, the operation, whether the recommendation was marked as carried out, and whether a later report found the same problem. Keep to management findings already made public; leave out anything on routes, stock levels or readiness.
Why it is still open
GAO's 2003 Iraq report says problems it found in the 1991 Gulf War had recurred, but each audit stands alone and readers must connect them by hand. A search found theses that trace recurring problems in prose, but no coded table across operations or countries.
Built with what exists
GAO reports and its public recommendation-status pages; NAO and ANAO report archives; pdftotext; an AI model such as Claude filling a fixed form for each finding with a quoted sentence as evidence (see the Note "Seven more flagged"); a spreadsheet; GitHub.
First step
Code the findings in GAO's 1992 Desert Storm redeployment report and its 2003 Iraq report against a first list of 10 problem types, and publish the coding rules.
Done looks like
At least 300 findings from at least three audit offices, 1991 to 2025, each linked to its page; two people independently code a random 60, with their agreement published as Cohen's kappa (a standard agreement score) of at least 0.7; and a count of problem types that recur in three or more operations.