Summary

  • Reporting on the U.S. Army’s Project Convergence exercise at Fort Irwin, in which 9,700 infantry soldiers fought a Blackhorse unit standing in for China’s PLA over 10 days in the Mojave Desert, presents a pattern in which products validated against the conditions of their use performed, while products that had not been so validated struggled.
  • The reported failures concentrate in three named programs — an Anduril drone forced to land in desert heat, a Theiss UAV that took roughly 10 minutes to launch against a Ukrainian benchmark of under one minute, and augmented-reality headsets whose deficiencies left the service “uncertain whether it would move forward with them” — each tied in the substrate to specific qualification gaps.
  • The reported gains — a Saronic unmanned boat used to rescue a stranded helicopter crew in the conflict with Iran, a Shahed-derived “Lucas” drone priced at $10,000–$55,000 per unit, Starlink-enabled long-range radio, and the Next-Generation Command and Control system with a $4 billion price tag in the current budget — correspond in the source to systems already employed in the field.
  • Competing diagnoses — environmental brittleness, procurement-model friction, doctrinal and training lag, institutional lag, expected experimentation — remain individually consistent with the failures but cannot be cleanly separated on the substrate, which anchors the explanatory weight in engineering condition and qualification history rather than in any single dominant cause.

The U.S. Army’s July Project Convergence exercise at Fort Irwin — 9,700 infantry soldiers against a Blackhorse unit standing in for China’s PLA over 10 days in the Mojave Desert — produced an uneven record that the source ties directly to whether individual systems had been validated against their operating conditions before being put in soldiers’ hands. Where the source documents combat or operational use (a Saronic unmanned boat employed in the conflict with Iran, a Shahed-derived “Lucas” drone described as “among the military’s most successful low-cost offensive weapons,” Starlink connectivity that allowed Lt. Col. Shawn Scott, “for the first time in his 19-year Army career,” to radio without cumbersome transmission poles, and the Next-Generation Command and Control system cutting mission planning time by two-thirds), the reporting describes performance. Where the source documents qualification gaps (an Anduril drone “previously tested in hot and cold temperatures” that nonetheless could not loiter in Mojave glare; Theiss UAV airframes “more than three years old” with batteries that “may not have been at peak performance”; AR headsets laid out “on a hot rock in 109 degree temperatures”), the reporting describes failure. The substrate carries both halves and does not adjudicate between them; the engineering-condition reading sits alongside the procurement-strategy, doctrinal, institutional, and expected-experimentation readings without resolving which dominates.

The substrate pattern: validation history as the consolidating variable

The reporting’s organizing variable, on close reading, is engineering condition — specifically, whether each named system had been qualified against the environmental and operational profile of its intended use before soldiers touched it. The substrate supports this reading on the failures (one Anduril drone tested in hot and cold temperatures that the Mojave heat forced to land; Theiss airframes more than three years old with batteries that “may not have been at peak performance”; AR headsets put on hot rock in 109-degree temperatures) and on the reported gains alike, though it does not establish formal qualification documentation for any of the systems that performed.

Whether each reported gain was formally qualified against its operational environment before Army deployment cannot be confirmed from the source. The source describes use cases and outcomes for Saronic, Lucas, Starlink, and NGC2 — combat employment in the Iran theater, public-sector demonstration of cost and operational tempo, connectivity already in routine use — but does not reproduce qualification-test records or environmental-testing artifacts. The validation reading therefore rests on the source’s reported employment history, not on formal qualification documentation, and that gap is itself a finding.

Failures at the fielding interface

Anduril drone

Sgt. Ramontez Bennett’s Anduril drone “was supposed to loiter above the action”; the Mojave glare forced it to land. Bennett, quoted in the source: “They gotta do something about it overheating.” An Anduril spokesman said the company’s drone overheated and noted it had previously been tested in hot and cold temperatures; following soldiers’ feedback, the company has prioritized upgrades for extreme environments, the spokesman said. The two statements are consistent: the drone was tested, and it failed at Fort Irwin. Wind compounded the issue: “All it takes is one gust of wind and it’s on the side of a mountain,” Bennett said; behind the rugged peaks, drones lost their signal.

Theiss UAV

A Theiss UAV next to Bennett’s station required “a prolonged wait to charge a drained battery” before launching, with the aircraft taking “about 10 minutes to get airborne.” Ukrainian soldiers, the source reports, “can get a drone of about the same size off the ground in under a minute.” Theiss founder Shawn Theiss attributed the delays to “inadequate soldier training or … glitches in communications parts made by a separate company,” and noted the drones at the exercise “were more than three years old” and that batteries “may not have been at peak performance.” Theiss said his drones typically launch within three minutes and that he received mostly positive feedback from soldiers at the exercise. Whether the delay was caused by training, third-party components, or equipment age is not resolved in the source.

AR headsets and the heat envelope

A senior defense official at the training event was quoted as saying the Army found during the exercise that the augmented-reality headsets referenced in the media invitation to Fort Irwin “showed deficiencies,” and that the service “is uncertain whether it would move forward with them.” Three companies have won AR headset contracts each worth more than $100 million. Lt. Gen. Michael McCurry, one of the senior Army leaders running the event, framed the failure mode in environmental terms: “You take a piece of electronics and lay it out on a hot rock in 109 degree temperatures and try and make it work. We’re seeing some of that.” The substantive point — that systems unqualified for the heat envelope fail when exposed to it — applies across vendors and device classes.

Convergence as high-diagnosticity evidence

The source’s broader environmental reporting reinforces the cluster reading: “servers overheated, batteries died and could not be recharged,” antennas “were on the fritz,” and “soldiers drenched T-shirts in ice water and draped them over” equipment. The convergence of failures across multiple vendors and multiple device classes in the same thermal environment registers as high-diagnosticity evidence for brittleness at the fielding interface rather than at any single vendor’s design. Whether each vendor had qualified its product against the same temperature profile before contract award cannot be confirmed from the source.

Reported gains — combat and operational validation

Saronic unmanned boat and the Iran conflict

The source reports that, during the conflict with Iran, “U.S. forces used an unmanned boat built by defense startup Saronic to rescue a stranded helicopter crew.” The Iran conflict “has relied largely on older and expensive systems,” per the source, framing the Saronic deployment as a contrast within an otherwise legacy force. Whether Saronic’s boat had been formally qualified for the recovery mission before deployment is not stated.

Lucas drone and Shahed-derived design

A drone modeled on Iran’s Shahed, “called Lucas,” has been “among the military’s most successful low-cost offensive weapons,” costing “$10,000 and $55,000 per unit through an arrangement in which the Pentagon owns the design and farms out production to a group of manufacturers.” The cost figure and the production-sharing arrangement are reported as a deliberate departure from conventional procurement; the source does not specify the qualification path that placed the design under Pentagon ownership.

“Starlink connectivity has allowed Lt. Col. Shawn Scott, for the first time in his 19-year Army career, to radio a soldier sitting in an air-conditioned operations center at another end of the base without cumbersome transmission poles.” “New data-sharing technology has cut by two-thirds the time it takes Scott to plan and communicate missions.” Starlink’s reported deployment at Fort Irwin reflects routine use rather than novel integration; the reported gain here is the elimination of an old workaround, not the introduction of a new capability.

NGC2 and dispersed operations

The Army’s Next-Generation Command and Control system, “with a $4 billion price tag in this year’s budget,” “allows troops to move faster and in a more dispersed fashion over longer stretches of battlefield — a capability that officials say is essential in modern conflicts.” The source describes NGC2 as “a start” — a characterization that qualifies the gain as partial and as still requiring complementary reliability upgrades.

Doctrinal and operational behavior

Soldiers at the exercise “flew the drones during limited windows and at roughly half the maximum flight time to conserve battery life and avoid losing them, operating without full autonomy or any AI to assist.” The source notes “Chinese consumer drones could do the manual job a decade ago.” A swarm of five small drones “was forced to turn back as three Black Hawk helicopters carrying senior Army officials flew into the same airspace, squandering a rare high-tech moment the visitors had come to see.”

Two readings sit side by side here. The first: operators had not yet internalized the new systems as primary tools, suggesting doctrinal lag. The second: the exercise was “designed to let specific units experiment with drones during pockets of fighting and learn about the technology — not to flood the airspace with autonomous weapons,” as an Army spokesman for Project Convergence said — suggesting some of the conservative behavior was prescribed rather than improvised. Brig. Gen. Daniel Hibner, commander of the Army’s Joint Modernization Command, made the doctrinal point explicitly: “It’d be a very dangerous time for us just to hand over everything to technology. … It’s very much still a human endeavor.”

The behavioral reading is therefore diagnostic of an adoption curve that is still climbing, while preserving the ambiguity that the curve was meant to be climbed slowly at Fort Irwin.

Competing diagnoses

Procurement-model friction

Defense Secretary Pete Hegseth “issued a directive to transform the Army into an AI-powered force equipped with drones and 3-D printers.” Per the source, he “instructed the Defense Department to stop looking for technological perfection and instead buy off-the-shelf products that are ‘good enough,’” stressing urgency. The procurement-model reading holds that off-the-shelf buying produces equipment meeting specifications but lacking ruggedization, integration testing, and lifecycle refresh of purpose-built systems. The reading is consistent with the failures but cannot be cleanly distinguished from brittleness on the available record. Theiss’s rebuttal, which shifts the locus onto training, third-party components, and equipment age, leaves the procurement-strategy reading intact without resolving it.

Doctrinal and training lag

The flight-time management and manual operation under battery-conservation pressure fit a doctrinal-and-training-lag reading. The Ukrainian benchmark — drones off the ground in under a minute — anchors the comparison: Ukrainian forces have iterated drone employment under combat loss for more than three years, refining acquisition and tactics together. The U.S. exercise, by design, has not. The doctrinal-lag reading is consistent with the source but is complicated by the prescribed-pacing counter-reading above.

Institutional lag

Daniel Goure (Lexington Institute, with war-college and Pentagon background), in the source: the Army’s history of modernization efforts is “almost unblemished by success,” attributed to “a tradition of risk aversion and a hierarchical organization.” Recently ousted Army Secretary Dan Driscoll characterized the institution as a “calcified bureaucracy”; Driscoll departed in September “following disagreements with Hegseth.” Jonathan Rue (former deputy assistant secretary of defense, partner at MVP Ventures): “The tech is moving so quickly that the U.S. military bureaucracy cannot keep up.” Congressional probes and oversight reports have found, per the source, that the service “has wasted billions of taxpayer dollars on fumbled new weapons programs and has suffered from inadequate training and failures in cybersecurity and data management.” “During battlefield drills held in Germany earlier this year, Ukraine’s drone operators took out a U.S. armored brigade.” The institutional account is consistent with the failures but does not, on its own, explain why commercial off-the-shelf equipment failed at the fielding interface.

Expected experimentation

Alex Miller (chief technology officer to the Army’s chief of staff): the event was meant to “fact-check all the vendors that are coming to us” about the veracity of their sales pitches. An Army spokesman for Project Convergence: the exercise was designed to “let specific units experiment with drones during pockets of fighting and learn about the technology — not to flood the airspace with autonomous weapons.” The fact-check framing originates with Army officials whose own communications described the technology as experimental, even as the same technology had won sizable contracts. Per the source: “much of the technology had already won sizable Army contracts, officials and soldiers involved described it as experimental rather than essential new weaponry.” The reading carries diagnosticity for the failure pattern, but the source does not adjudicate whether the contracts preceded fielding (procurement outran testing) or whether contract award was itself exploratory (in which case the fact-check framing holds).

The Ukraine comparison

The Ukraine comparison is diagnostic of both an adoption-tempo gap and an employment-doctrine gap: Ukrainian forces have iterated drone employment under combat loss, refining acquisition and tactics together. Per the source: “The Ukraine war brought the inadequacies into focus, putting fully autonomous warfare on the battlefield with a technical proficiency that has at times humbled the U.S.” The comparison is less diagnostic of a procurement-strategy failure as such, because the source does not specify whether U.S. drones in question were procured through the same channels or designed to the same specifications as Ukraine’s drones. The strongest substrate-anchored reading: the comparison shows what field iteration under pressure produces (in acquisition and in employment doctrine together), and what an exercise designed to let “specific units experiment with drones during pockets of fighting” is not designed to produce.

Who has standing in the Fort Irwin record

Definitive stakeholders

  • Office of the Secretary of Defense. Urgency through Hegseth’s transformation directive; legitimacy through statutory authority; power through the directive’s reach into acquisition policy.
  • Army senior leadership. Urgency through running of the event (McCurry, Hibner, Merkl); legitimacy through chain-of-command role at Fort Irwin; power through resource control.
  • Congress. Power and legitimacy through long-pressed modernization oversight and budgetary authority; per the source, “Congress has long pressed the Army to modernize.”
  • Operational formations. Urgency through reference to “a technologically sophisticated adversary in China.”

Dependent stakeholders

  • Soldiers at Fort Irwin. Flight-time management and manual operation under battery-conservation pressure embodied the doctrinal-lag hypothesis in real time.
  • Ukrainian military. Battlefield performance sets the implicit benchmark cited in the source — “Ukraine’s drone operators took out a U.S. armored brigade” during German drills earlier in the year.
  • Defense startups. Per the source, soldiers said their technology “has at times proven too brittle for real-world applications.” The source does not directly address the relative procurement-urgency of startup versus Army adoption cycles.

Dormant or discretionary stakeholders

  • PLA (notional adversary). Held at high power and urgency; operates outside the exercise’s accountability frame.
  • Taxpayers. Diffuse legitimacy claim, not directly represented in the procurement chain.
  • Future Army formations and allied interoperability partners. Interests created by current decisions; not represented in present procurement deliberations.

What happens next

Immediate horizon

The Army is likely to continue incremental fielding under the “good enough” model and to run additional exercises surfacing further gaps. Congress will continue oversight pressure.

Medium term — conditional on dominant driver

If environmental brittleness is dominant, the implication is either an upgrade cycle that erodes the cost advantage of off-the-shelf procurement or doctrinal adaptation accepting lower reliability for fielding speed. If institutional lag is dominant, the relevant intervention is procurement reform, not hardware redesign. The source does not establish which dominates.

Long term

Per the source, “the last major Army transformation came in the 1980s, when the service pivoted from Vietnam and built the force it would use in the Gulf War — retraining, reorganizing and fielding armored fighting vehicles, attack helicopters and modern air defense systems.” The current effort faces a peer-adversary timeline that, per the source’s reporting, may not permit comparable pacing.

Next-determining observations

Whether the three AR headset contracts are renewed; whether Anduril and Theiss platforms reappear at Fort Irwin or successor exercises with revised qualification documentation; whether Saronic, Lucas, Starlink, and NGC2 are documented as formally qualified against their operational environments. Per the source: “Some defense officials said the efforts would bear fruit longer term despite near-term stumbles.” The substrate does not establish whether that forecast is correct.

How this is being framed

Army’s own framing

Merkl: “we are absolutely dedicated to transformation.” Miller (fact-check framing of the exercise). Hibner: warfare is “very much still a human endeavor.” Army spokesman for Project Convergence: “pockets of fighting” experimentation. The leadership positioning is consistent across three of the principals quoted on the program’s purpose.

Surfaced tension within the analytical record

One reading commits to a unifying thesis: the diagnostic signature at Fort Irwin is consistent with a procurement-validation gap concentrated in unqualified programs, and inconsistent with a uniform institutional-failure reading of every Army modernization line item. A second reading preserves multiple candidate diagnoses (environmental brittleness, procurement-model friction, doctrinal and training lag, institutional lag, expected experimentation) and does not adjudicate, holding the Fort Irwin shortfalls’ deeper cause unresolved on the available record. Both anchor in the same substrate. The convergence is on engineering condition as the substrate-anchored explanation for the failure cluster; the divergence is on whether to commit to a single dominant diagnosis.

Procurement-versus-framing coherence tension

Three companies had already won contracts each exceeding $100 million for headsets whose deficiencies, per a senior defense official at the training event, leave the service “uncertain whether it would move forward with them.” The source does not resolve whether contracting preceded fielding (implying procurement outran testing) or whether contract award was exploratory (in which case the fact-check framing holds). The two readings imply different procurement reforms.

What the source establishes and does not establish

Quotes reproduced verbatim from the source

McCurry (“You take a piece of electronics and lay it out on a hot rock in 109 degree temperatures and try and make it work. We’re seeing some of that”); the senior defense official quoted on headset uncertainty; Merkl (“there is a recognition that the battlefield is changing before our eyes”; “we are absolutely dedicated to transformation”); Hibner (“It’d be a very dangerous time for us just to hand over everything to technology”; “very much still a human endeavor”); Goure (“almost unblemished by success”; “a tradition of risk aversion and a hierarchical organization”); Rue (“the tech is moving so quickly that the U.S. military bureaucracy cannot keep up”); Miller (“fact-check all the vendors that are coming to us”); Theiss on age and battery condition; the Army spokesman on “pockets of fighting” experimentation.

Reported statements carried without inner quotation marks

The Anduril spokesman’s claims regarding prior testing and upgrade priorities; the descriptions of Starlink and NGC2 performance; the descriptions of the soldiers’ drone-handling behavior. The source reports the Anduril statement indirectly rather than as a direct quote; the exact wording attributed to the spokesperson is not in the substrate.

Hedges preserved

“may not have been at peak performance”; “reported”; “said”; “according to” — where present in the source, preserved in this Analysis.

Symmetric application

Substrate failures and reported gains carried together; stakeholder mapping applied symmetrically across coalition groupings; no coalition-relativized standard.

Attribution discipline

Pattern-level attributions carried to quoted sources (Goure on institutional lag; Rue on bureaucratic tempo; Hibner on the human-endeavor framing; Miller on fact-checking; Theiss on equipment age and training). Aggregate characterizations stay anchored to specific source statements.

Diagnostic-frame attribution

Diagnosticity ratings and the consolidating reading (engineering condition and qualification history as the consolidating variable) carry an explicit methodological operator — they are judgments under the differential-diagnostic and substrate-anchoring frames, not source claims.

Gaps the substrate does not establish

  • Qualification history of named gains. Whether Saronic, Lucas, Starlink, and NGC2 were formally qualified against their operational environments before Army deployment cannot be confirmed from the source; the source describes use cases and outcomes, not formal qualification documentation.
  • Pattern vs. selection. Whether the failure-success distribution reported reflects a substrate pattern or a narrative-selection pattern within the WSJ article cannot be determined from a single article alone; cross-referencing Army program-status documents would be needed to verify.
  • Procurement-urgency comparison. The relative procurement-urgency of defense-startup versus Army adoption cycles is not directly addressed by the source.
  • Exercise purpose adjudication. The fact-checking framing is described but not externally validated as a complete account of the event’s purpose.
  • Anduril spokesperson phrasing. The source reports the Anduril statement indirectly rather than as a direct quote; the exact wording attributed to the spokesperson is not in the substrate.

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

Coherence Audit
Tests whether an argument hangs together — spotting contradictions, gaps, and circular reasoning.
Differential Diagnosis
Lists the candidate explanations for a symptom and rules them out one by one.
Stakeholder Mapping
Charts the parties to a situation — their interests, power, and alignments.