Owen Gregorian@OwenGregorian
Army AR Goggles: GAO Confirms $1.8B Spent on Headsets Too Sick to Field | Mireya Ramsey, Techtimes
Congress spent $1.8 billion on nearly 10,000 Army augmented-reality combat goggles that will now sit in storage rather than reach soldiers — a conclusion delivered this week by federal watchdogs who found the program produced headaches, nausea, and worse battlefield performance than the equipment it was meant to replace. The specific engineering failure is documented and quantifiable: the goggles' 70-degree display field of view, combined with HoloLens 2 display latency that the underlying optics could not resolve, triggered vestibular conflict severe enough that more than 80% of soldiers reported physical symptoms within three hours of wearing the device.
The Government Accountability Office's June 2026 report characterized the Army's Integrated Visual Augmentation System (IVAS) as a program that produced "nearly 10,000 units of the first two versions" that "have fallen short of soldiers' needs and will go into storage, with some potentially used for testing, rather than to the field." The verdict was reinforced at a House hearing Tuesday, where Carmen Malone, an assistant inspector general who audits Defense Department acquisition programs, traced the failure to overambition baked in from the start.
Soldiers Performed Worse With IVAS Than Without It
The IVAS concept was designed to give infantry soldiers a mixed-reality heads-up display capable of overlaying sensor data, terrain maps, targeting information, and drone feeds directly onto their field of vision. When the Army Acquisition Executive formally approved development in September 2018 and Microsoft received an initial contract the following November, the program's pitch was audacious: 120,000 headsets across the close combat force, with Microsoft claiming the system would allow soldiers to "see through smoke and around corners" and have three-dimensional terrain maps projected onto their field of vision in real time.
The gap between that promise and what the hardware could actually deliver emerged within months of fielding. The Department of Defense's own testing office ran early IVAS 1.0 units through training scenarios in 2022 and found that soldiers hit fewer targets and engaged them more slowly than with their existing equipment, according to the DOT&E FY2022 annual report. Soldiers reported feeling disoriented. They developed eye strain, neck strain, and nausea — what the Pentagon's testing office formally classified as "mission-affecting physical impairments." One soldier's assessment, which became widely cited, was direct: the devices "would have gotten us killed."
The root cause was not merely that the hardware was uncomfortable. The HoloLens 2 platform, which was never designed for military use, used waveguide optics targeting a 70-degree horizontal field of view — more than double the 40-degree field of legacy night-vision goggles that soldiers had used for decades. To achieve that wider field, the display system required stacking multiple waveguide panels, which increased weight, degraded brightness in outdoor lighting conditions, and introduced display latency above the threshold at which the vestibular system begins to detect incoherence between perceived and actual motion. That conflict is cybersickness — the same mechanism that causes nausea in commercial VR headsets, scaled up into a sustained operational environment where soldiers were running, climbing, and clearing rooms.
A 2022 audit by the DoD Inspector General found that the program had proceeded to production without ever defining "minimum user acceptance levels to determine whether IVAS would meet user needs." That gap, the IG warned, risked wasting the full $21.88 billion contract value on a system soldiers would refuse to use. The Army contested those conclusions as "fundamentally flawed and inflammatory" — then bought 5,000 more headsets.
Why the Army Bought 10,000 Headsets It Knew Didn't Work
The IVAS program procured 5,000 IVAS 1.0 systems in 2022 and another 5,000 IVAS 1.1 systems in 2023. Both decisions were made before either version had been validated through operational testing in the field. The GAO's report explicitly noted that "after negative soldier feedback on version 1.0, the Army didn't even conduct operational testing on version 1.1 before production." That sequence — buying before testing — is the specific manifestation of what the GAO calls "failing slow."
The structural reason for that sequence is documented in the same GAO report. The Pentagon's budget process requires programs to secure long-range funding commitments before their business cases are fully understood. Success within that system is measured by how much money has been obligated, not by whether capability has been delivered. Obligating funds rapidly protects budget allocations from being redirected to other programs during the congressional appropriations process — so program managers have a structural incentive to buy units before testing, not after.
"When requirements are unstable or overly ambitious," Malone told the House subcommittee this week, "programs pursue systems they are not ready to build." That is an accurate description of IVAS's trajectory, but it is also a description of the category of failure — not its cause. The cause is that the program's managers faced institutional incentives that rewarded moving money faster than moving technology forward. The GAO has documented this pattern continuously since 1990.
By June 2026, Army spokesperson Ellen Lovett confirmed to Task & Purpose what multiple years of testing had established: IVAS 1.2, the most advanced version that reached prototype stage, was "unaffordable to produce at scale" and the Army had "decided to pivot" to a new program. Unit costs had escalated from an initial estimate of roughly $29,000 per headset to approximately $80,000 per unit — a figure the Army acknowledged was incompatible with fielding at the scale of 100,000-plus soldiers.
How EagleEye Plans to Solve What HoloLens Could Not
The program's industrial architecture changed substantially before the final verdict arrived. In February 2025, Microsoft announced it would partner with Anduril Industries, ceding "oversight of production, future development of hardware and software, and delivery timelines." By April 2025, a formal contract novation transferred Microsoft's IVAS agreement to Anduril. By September 2025, the Army had rebranded the successor effort as Soldier Borne Mission Command (SBMC) and awarded two competing hardware prototype contracts: $159 million to Anduril for its EagleEye headset system, and $195 million to startup Rivet — whose CEO, Dave Marra, had previously served as IVAS program director at Microsoft.
Anduril's EagleEye, unveiled at the Association of the United States Army's annual meeting in October 2025, takes a fundamentally different architectural approach than IVAS. Where IVAS adapted a consumer headset to a military helmet, EagleEye was designed from the start as a helmet-integrated system. The battery, compute hardware, and radio — the components responsible for most of IVAS's helmet weight — are moved into a ballistic chest plate rather than mounted above the soldier's center of gravity. The result shifts the system's mass below the shoulder joint, eliminating the lever-arm torque on the neck vertebrae that IVAS imposed. Anduril's general manager for warfighter systems, Quay Barnett, described the ergonomic improvement in concrete terms at AUSA: "We've got this already optimized where it sits perfectly over the spine as your center of gravity."
The display and optics technology for EagleEye comes from Meta — the company that acquired Oculus VR in 2014 and later, in 2017, parted ways with Oculus founder Palmer Luckey. Qualcomm is providing the embedded processors for on-device AI inference. Gentex's Ops-Core division is handling the ballistic protection, and Oakley Standard Issue is handling ruggedization. The partnership is Anduril's argument that maturing commercial AR optics — built for Meta's Quest headset line and then adapted upward to military specifications — can outperform the inverse approach that IVAS took, which was building down from consumer hardware.
The software backbone is Anduril's Lattice platform — an AI-based sensor fusion system that integrates thermal imaging, low-light cameras, GPS, squad radio, and unmanned vehicle feeds into a unified battlefield picture displayed in the visor. Lattice also introduces large language model interfaces, currently being tested with multiple AI providers, that would allow soldiers to issue voice commands to autonomous systems directly from the headset. That capability did not exist in IVAS.
SBMC Faces the Same Structural Trap IVAS Did
The architectural improvements in EagleEye address the specific engineering failures of IVAS. Whether they are sufficient is an open question that only field testing will resolve — and field testing has not yet occurred. As of July 2026, no independent evaluation of EagleEye's performance, ergonomics, or long-term health effects has been published. The approximately 100 EagleEye-based SBMC prototypes that Anduril planned to deliver in the April-June 2026 window for soldier evaluation have not been publicly confirmed as delivered.
More significantly, EagleEye does not address the unresolved technical problem that undermined IVAS's mixed-reality training capabilities: dynamic occlusion. In a mixed-reality environment, the display must be able to render virtual objects as spatially "behind" real physical objects — walls, doors, other soldiers. Current waveguide-based AR displays cannot do this in real time without camera-based depth mapping that exceeds what current embedded sensors can provide. Anduril has not publicly stated whether EagleEye has solved this problem.
There is also the structural problem that has nothing to do with the hardware. The GAO's report that buried IVAS was written as the conclusion of one chapter, not the entirety of the story. The same structural incentive — obligating funds before capability is demonstrated, to protect budget allocations from congressional reallocation — will face SBMC program managers exactly as it faced their predecessors. The GAO has noted that a November 2025 DoD acquisition reform memorandum pledged to adopt iterative development and scale investment only as programs demonstrate progress. The watchdog also noted it intends to publish a separate fall 2026 report evaluating whether those pledges have actually changed the incentive structure. That verdict is not yet in.
"Waste in these programs occurs when they are structured to 'fail slow,'" the GAO's June 2026 report stated. SBMC will succeed or fail on whether the Army applies the same logic to its own procurement decisions when the first round of EagleEye soldier evaluations returns negative feedback — rather than buying 5,000 units and hoping the next version fixes it.
Anduril is targeting first scaled delivery of SBMC hardware in 2027. The Army is expected to choose between Anduril and Rivet for the production contract around 2028, if it chooses either.
Frequently Asked Questions
Why did the Army's IVAS AR goggles make soldiers sick?
The goggles produced cybersickness — a form of motion sickness triggered by a mismatch between what the visual system perceives and what the vestibular system (inner ear) detects. IVAS targeted a 70-degree horizontal field of view, more than double the 40-degree field of standard night-vision equipment. Achieving that wide a field required a waveguide display architecture that introduced display latency — a delay between head movement and display update — above the threshold at which the vestibular system registers incoherence. The result, confirmed in operational tests by 2022, was headaches, eyestrain, and nausea in the majority of soldiers who wore the headsets for more than three hours. The Army reduced the field of view to 60 degrees in IVAS 1.2, but that adjustment neither eliminated the underlying latency problem nor fully resolved soldier symptoms.
What is Anduril EagleEye and how is it different from IVAS?
EagleEye is a family of augmented-reality headsets developed by Anduril Industries as the primary hardware offering for the Army's successor program, Soldier Borne Mission Command (SBMC). Unlike IVAS, which adapted Microsoft's consumer HoloLens 2 headset to military helmets, EagleEye was designed from the ground up as a helmet-integrated system. Its battery, compute hardware, and radio are housed in a ballistic chest plate rather than mounted above the soldier's head, redistributing the system's weight to below the shoulder joint and eliminating the neck-strain problem that IVAS created. Anduril is developing EagleEye in partnership with Meta (display and waveguide optics), Qualcomm (processors), Gentex/Ops-Core (ballistics), and Oakley Standard Issue (ruggedization). No independent evaluation of EagleEye's field performance has been published as of July 2026.
Does SBMC face the same structural procurement risks that caused IVAS to fail?
The IVAS program failed partly because of display engineering problems, but also because the Army's budget process incentivized buying thousands of headsets before either version had passed operational testing — locking in expenditure before capability was validated. That structural incentive — obligating funds quickly to protect budget allocations — is not an artifact of the IVAS program specifically; it is a documented pattern in Pentagon procurement that the GAO has placed on its High Risk List since 1990. The November 2025 DoD acquisition reform initiative pledged to change this by linking investment scaling to demonstrated capability milestones. Whether SBMC program managers will actually follow that framework — or buy prototype units at scale before field evaluations are complete — is the test that will determine whether the Army's second attempt at combat AR learns from its first.
Why did it cost $1.8 billion and produce nothing soldiers can use?
The $1.8 billion figure reflects the cumulative cost of three hardware iterations between 2018 and 2025, none of which delivered a system soldiers were willing to use in the field. The initial HoloLens 2-based design was not ruggedized for military use and could not function in rain. Subsequent versions addressed ruggedization but not the fundamental vestibular conflict caused by display latency and wide FOV. Unit costs escalated from a planned $29,000 per headset to approximately $80,000 per unit by 2026, at which point the Army acknowledged the technology was "unaffordable to produce at scale." The approximately 10,000 units that were purchased will be stored rather than fielded; 400 IVAS 1.2 prototypes are being repurposed to inform SBMC's design process, according to the GAO's June 2026 report.
techtimes.com/articles/32016…
