DASNet, the Afghan Mission Network, and what the Taliba...

When a partner government collapses in days, the question is not only what weapons it leaves behind. It is what data, what identity records, and what communications infrastructure switch hands with it. In Afghanistan in August 2021, all three did. This post traces what the coalition actually built for the Afghan forces, what it cost from primary sources, and what became of it, keeping three very different dollar accountings rigidly separate along the way.

It also starts with a correction, because the honest version of this story requires one.

A name that does not map to a network

This entry was assigned around the term "DASNet." So it is worth stating plainly at the top: there is no reliable public source describing a coalition or Afghan National Defense and Security Forces (ANDSF) network literally named "DASNet." Searches surface only one confirmable referent, and it is a company, not a network.

DasNet Corporation is a network-systems-integration firm based in Bohemia, New York, that has provided command, control, communications, computers, and intelligence (C4I) systems engineering to the Department of Defense, the State Department, civilian agencies, and foreign governments, per the company's own description (a single-source, self-published claim). In September 2005, DasNet, teamed with Globecomm Systems, was awarded a contract to build an early communications link for the Afghan National Army. According to Via Satellite's contemporaneous reporting, the contract carried a base value of roughly $7.4 million plus about $3.6 million in options, for a potential total near $11 million, fully funded by the U.S. government. It provided equipment, personnel, and one year of operations and maintenance to connect the Afghan Ministry of Defense with the Afghan National Army.

That is the real DasNet story: a small contractor-built link between a ministry and its army in 2005. It is not a coalition-wide network, and it is not the thing most people are reaching for when they invoke a grand Afghan digital backbone. So rather than invent a definition, this piece anchors on what is documented: the coalition's actual information-sharing network, the Afghan Mission Network, and the biometric and personnel systems built for the ANDSF that later fell under Taliban control.

Keep that first dollar figure parked. The DasNet and Globecomm contract at roughly $7.4 million base and about $11 million with options is a rounding error against everything that follows, and it should never be blended into the larger totals.

The coalition network that actually worked: the Afghan Mission Network

If the term "DASNet" points at nothing, the Afghan Mission Network (AMN) points at a genuine, well-documented achievement.

AMN was the primary coalition C4ISR and information-sharing network for the International Security Assistance Force (ISAF). It was formally established in January 2010 by order of the ISAF Joint Command, under a directive titled "Migration to the Afghanistan Mission Network," and RAND's dedicated study (report RR302) corroborates both the date and the role. Its founding purpose was a posture shift that had defeated every prior coalition operation: moving from "need to know" to "need to share."

The hard technical problem AMN solved was federation. Dozens of nations arrived in Afghanistan with incompatible classified networks that could not talk to each other. AMN built an ISAF Secret core on top of the U.S. CENTRIXS-ISAF (CX-I) backbone and then let national networks plug in as extensions: the United Kingdom contributed OVERTASK, Canada contributed LCSS, Italy contributed CAESARNet, and so on. This distinction matters and is easy to get wrong. CENTRIXS-ISAF is the U.S. component and backbone. AMN is the multinational network-of-networks federated on top of it. They are not the same thing.

The scale it reached was substantial. Commonly cited figures put AMN at roughly 90,000 users by May 2011 and 48 NATO and partner nations operating on it by the end of that year. Those counts come from tertiary sources and vary by date, so treat them as approximate rather than precise. But the direction is not in doubt: this was a large, working, multinational network.

Its most durable legacy is doctrinal. RAND RR302, titled "Lessons Learned from the Afghan Mission Network: Developing a Coalition Contingency Network," documents how AMN's approach informed NATO's Federated Mission Networking (FMN) initiative and the U.S. Mission Partner Environment (MPE), frameworks still in use for coalition operations today. That is real public value: a contingency network whose design outlived the war it was built for and became a template for the next one.

The systems built for the Afghan forces, and the exit that failed

The second story is harder. Alongside the coalition network, the United States helped build layered digital identity and pay infrastructure for the ANDSF itself. These systems had defensible purposes. They also became a liability the moment the government they served collapsed. Three things are involved, and they are genuinely distinct.

HIIDE (Handheld Interagency Identity Detection Equipment) are field collection devices. They capture iris scans, fingerprints, and facial images. As the Taliban advanced in mid-August 2021, HIIDE units were physically seized, as reported by The Intercept, sourced to a Joint Special Operations Command official and three former military members. The precision point here is important and frequently overstated: a captured handheld alone gives only limited local access. The bulk of the biometric data resides on remote secure servers. A seized HIIDE device is not equivalent to full access to the underlying databases, and it should not be described as if it were.

AABIS (Afghan Automated Biometric Identification System) is the larger concern. It is the Ministry of Interior's biometric identity database, run out of a "Biometrics Center" and formally established around late 2009, modeled on the classified DoD ABIS system. Its stated aim was to keep Taliban fighters and criminals out of the army and police, and program planning reportedly targeted coverage of about 80 percent of the population by 2012. MIT Technology Review reported that experts view the databases, not the handhelds, as the greater threat. A widely repeated figure puts AABIS at roughly 8.1 million records, but MIT Technology Review explicitly flags that number as unconfirmed, traced to the LinkedIn profile of a former U.S.-based program manager. It should be attributed and hedged, never stated as an established fact.

APPS (Afghan Personnel and Pay System) is a personnel and pay system with biometric enrollment, launched around 2016 to curb payroll fraud. Per MIT Technology Review, it held roughly half a million records covering Afghan Army and Police members, with at least 40 data fields per person. Those fields went well beyond a name and a number: birth details, military specialty, the names of relatives, tribal-elder guarantors who vouched for each recruit, and trivia such as a person's favorite fruit or vegetable. The roughly 500,000 figure is an estimate from individuals familiar with the program.

Together, AABIS and APPS held exactly the kind of information that a counter-infiltration tool and a targeting list have in common: who served, where they live, and who their family members are.

What happened next is documented but must be stated precisely. Human Rights Watch reported in March 2022 that the Taliban control systems holding iris scans, fingerprints, photographs, occupations, home addresses, and relatives' names. HRW documented at least one former commander whose fingerprints and irises were scanned during detention. HRW is an advocacy organization, and that detention account is a single documented anecdote, not a quantified pattern. The honest characterization is a serious, documented concern with evidence of at least isolated use, not a proven case of systematic mass targeting at scale.

The money, kept in separate buckets

This is where the recurring traps live, so the accounting is deliberately compartmentalized.

Bucket one: the DasNet and Globecomm contract. Roughly $7.4 million base, about $11 million with options, 2005, per Via Satellite. A small, early ANA communications link. Nothing else in this story should absorb it.

Bucket two: system-specific costs and the sharpest oversight finding. This is the improper-payments case at the heart of the critique. A SIGAR audit released in July 2022 (report 22-28-IP) found that the Department of Defense spent $64.8 million building the APPS software, the system specifically meant to biometrically de-duplicate the force and stop "ghost soldier" fraud. Then its controls were bypassed. SIGAR found that $232 million in Ministry of Defense salary payments over the period from fiscal year 2019 to May 2021 went to "suspicious units and non-existent object codes," or were salaries that never reached real personnel bank accounts, as reported by Lawfare on the audit. The audit also flagged more than 7,000 duplicate identification records and over 1,000 invalid ones. Keep the two numbers distinct: $64.8 million is what the anti-fraud system cost to build; $232 million is what leaked around it.

Related, and worth stating carefully, is the "ghost soldier" problem. SIGAR repeatedly reported that ANDSF payroll numbers were inflated by non-existent personnel, and that APPS, though designed to de-duplicate the force biometrically, was never fully used to manage Ministry of Defense payroll. Specific hard counts of ghost soldiers, such as figures near 200,000, circulate in secondary press and are contested. SIGAR's defensible framing is not a precise ghost count but the more sobering fact that the true strength of the ANDSF was never reliably known.

Bucket three: macro totals, context only. The United States appropriated roughly $80.7 billion to the Afghanistan Security Forces Fund (ASFF), about 91 percent of all security-sector reconstruction appropriations, per the Congressional Research Service (report IN11728), with total security-sector appropriations near $88.6 billion. Some SIGAR tabulations cite the ASFF closer to $82.9 billion and the security sector near $88.8 billion; the figures drift by report and date, which is itself worth noting rather than papering over. Total U.S. Afghanistan reconstruction appropriations were about $148 billion from 2002 to 2021, roughly 60 percent security-related, and SIGAR's December 2025 final forensic report characterizes the overall effort as a large-scale failure. These totals cover equipment, salaries, training, fuel, and construction across two decades. Networks and biometrics are a small slice of that whole, and the macro figures must never be presented as the cost of any single system.

One more accounting sits adjacent and must also stay separate: the value of equipment left behind. DoD reported roughly $7.1 billion in ANDSF equipment abandoned in 2021, including tens of thousands of vehicles, hundreds of thousands of weapons, and over 160 aircraft, and SIGAR's final report describes that equipment as forming a "core" of the Taliban's post-2021 military. SIGAR itself cautions that the $7.1 billion figure may not be accurate because of gaps in DoD's inventory-management systems. That is an equipment-value accounting, not a program cost, and not the same as the $80 to 88 billion spent standing up the force.

The critique and the defense, side by side

The honest critique. The United States spent roughly $80 to 90 billion building the ANDSF, including identity and pay infrastructure meant to end corruption and infiltration, and much of that infrastructure either failed on its own terms or became a liability. The clearest single failure is the pay system: DoD built a $64.8 million tool to stop payroll fraud, then bypassed its controls while $232 million flowed to questionable payments, per SIGAR 22-28-IP. The identity systems succeeded technically and failed catastrophically at the exit. HIIDE devices were captured, and the AABIS and APPS databases fell under Taliban control, converting a counter-infiltration tool into a potential targeting resource pointed at the very Afghans who trusted the coalition. The oversight lesson is specific: building sensitive identity infrastructure without an equally serious plan for its custody, decommissioning, or destruction under conditions of collapse is itself a form of waste. The money bought capability that outlived the mission and switched sides.

The honest defense. These systems answered real wartime needs, and one of them was a genuine achievement. The Afghan Mission Network solved a problem that had defeated every prior coalition operation, letting dozens of nations with incompatible classified networks share a common operating picture, and it seeded FMN and MPE doctrine still in use, per RAND RR302. That is durable public value. The ANDSF-facing systems also had legitimate purpose. AABIS and APPS existed to keep infiltrators out of the force and to stop payroll fraud that was bleeding U.S. taxpayers, and biometric enrollment was a defensible way to know who was actually serving. The early DasNet and Globecomm link connected a ministry to an army that otherwise had no secure means to command it. The tragedy is not that the coalition built identity and networking infrastructure. That was reasonable. It is that no one adequately planned for what happens to sensitive data and equipment when a partner government collapses in days. The mission was real. The exit was not.

Fact-check notes and sources

DasNet Corporation is a C4I network-systems-integration firm (Bohemia, NY); this is the only confirmable referent for "DASNet," a company self-description and single source: dasnetcorp.com.
The DasNet and Globecomm ANA communications contract (September 2005, ~$7.4 million base, ~$3.6 million options, ~$11 million potential, U.S.-funded, MoD-to-ANA link): Via Satellite.
The Afghan Mission Network was established January 2010 by the ISAF Joint Command, federated the U.S. CENTRIXS-ISAF backbone with national extensions, and seeded FMN/MPE doctrine: RAND report RR302, rand.org, with architecture and scale corroborated at Wikipedia (tertiary; ~90,000 users and 48 nations presented as approximate).
HIIDE biometric devices were seized during the August 2021 collapse; handhelds alone give limited local access, not full database access: The Intercept.
AABIS (Ministry of Interior identity database, ~2009, modeled on DoD ABIS) and APPS (personnel/pay system, ~2016, ~500,000 records, 40+ data fields) distinctions, and the databases-over-devices risk; the ~8.1 million AABIS record count is flagged as unconfirmed and LinkedIn-sourced: MIT Technology Review.
Taliban control of the abandoned biometric systems and at least one documented detention case (advocacy source; single anecdote, not a quantified pattern): Human Rights Watch, March 2022.
$232 million in questionable MoD salary payments (FY2019 to May 2021) made outside APPS controls, against $64.8 million spent building the APPS software: SIGAR report 22-28-IP (July 2022), sigar.mil PDF, reporting by Lawfare.
ASFF ~$80.7 billion appropriated (~91% of security-sector appropriations; ~$88.6 billion total security sector; figures drift by report): CRS report IN11728, congress.gov PDF.
Total Afghanistan reconstruction appropriations ~$148 billion (roughly 60% security-related) and the abandoned-equipment "core" finding; SIGAR December 2025 final forensic report, which itself cautions the ~$7.1 billion equipment figure may be inaccurate due to DoD inventory gaps: SIGAR final report PDF.

DASNet, the Afghan Mission Network, and what the Taliban inherited