11 minute read

TL;DR: On August 7, the Army opened a request pathway to four stateside military test ranges and included a developing international venue in Morocco. Companies can submit without a prior federal contract or existing government relationship. That is real reform. It removes a barrier that has kept capable commercial teams from testing drones, counter-drone systems, long-range fires, low-cost interceptors, and related technologies under conditions they cannot reproduce on a commercial field. But access is only the first gate. It is infrastructure reform, not an acquisition decision. A range event creates fielding value when it starts with a falsifiable claim, uses representative conditions, captures decision-grade data, controls the tested configuration, and ends with a named transition decision. Otherwise, “fly, fail, fix, fly” becomes an expensive sequence of demonstrations. The Army opened the gate. Industry still has to bring the evidence.

This post uses public sources and a framework drawn from general professional experience. It contains no nonpublic program, platform, customer, employer, or source-selection information.

The Army did something unusually practical this month. It opened the range.

On August 7, the service announced that private companies could request access to some of its premier test environments through a single digital front door. A company does not need an existing federal contract or prior government relationship merely to submit a request.

That matters more than another innovation memo.

I have spent much of my career in and around open-air ranges, live and virtual training environments, and test programs. The land is not the rarest asset. The real capability is the combination of controlled airspace, spectrum, safety, instrumentation, threat representation, experienced people, and enough physical scale to let a system fail honestly.

Commercial developers need that capability. They also need to understand what it can and cannot do for them.

What the Army Actually Changed

The Army announcement names four stateside locations:

• Dugway Proving Ground in Utah;
• West Cibola Range, a newly opened component of Yuma Proving Grounds in Arizona;
• Camp Shelby Joint Forces Training Center in Mississippi; and
• Camp Grayling Joint Maneuver Training Center in Michigan.

The announcement also points to Morocco’s Multidomain Training Area, a component of the broader Africa Multidomain Training and Experimentation Center effort. The MDTA is already supporting experimentation, but that location needs a qualifier: AFRICOM says the full AMTEC is being established in Tan-Tan toward 2030, with African Lion 2027 serving as a proof of concept. It should be treated as a partner venue and developing international pathway, not simply another Army-owned commercial range.

The initial emphasis is on drones, counter-drone technologies, long-range fires, and low-cost interceptors. The Army also announced a special electromagnetic-reliability event at Camp Grayling during the week of September 13, with similar opportunities anticipated quarterly. That detail is important. A drone that works only in clean spectrum is not ready for the environment it was built to enter.

Industry can submit an initial request through the Army Range Reservation portal, where a concierge team will help applicants identify a suitable location and window. The announcement is equally clear about the constraint: submitting a request does not guarantee the desired range or schedule because capacity is finite.

This is an access pathway, not an entitlement, a free test program, or an acquisition award.

That is still a significant change.

Why Range Access Matters

Defense innovation has more than one valley of death.

There is the familiar transition gap between a successful prototype and funded procurement. Before that comes a quieter gap between something that works in a company-controlled environment and something that has been credibly tested under operationally relevant conditions.

The Department’s own APFIT program describes that lab-to-prototype gap: companies struggle to find the resources, test environments, and partners needed to prove real-world viability. The Army Applications Laboratory’s FY2025 performance report found range awareness and access was the second-largest innovation barrier it measured, behind reliable transition funding. It also found that 43 percent of its projects described as transitioned still experienced a follow-on contract gap longer than three months.

That is the boundary. Opening Army test infrastructure attacks the evidence bottleneck. It does not create the requirement, sponsor, funding line, or next contract.

A commercial field can show that a vehicle moves, a sensor detects, or an application runs. It usually cannot reproduce the airspace, terrain, spectrum congestion, safety controls, threat systems, instrumentation, and scale needed to establish where performance breaks.

That last question is the one test exists to answer.

The strongest systems are not the ones that never fail in a demonstration. They are the ones whose developers know the boundary, understand the failure, correct it, and can prove what changed.

A Demonstration Is Not a Test

A demonstration asks whether the system can perform a planned sequence under managed conditions. Defense Acquisition University guidance describes demonstration as observation when quantitative assurance is not required, while a test generates data under controlled and traceable conditions for quantitative evaluation.

A test asks whether a defined configuration achieves a measurable result under stated conditions, how repeatable that result is, and what happens when the conditions move outside the happy path.

Both have value. They are not interchangeable.

The distinction gets blurred because a good demonstration is visible. Leaders can watch it. Cameras can record it. A test produces less glamorous artifacts: synchronized data, ground truth, anomaly records, configuration baselines, confidence bounds, and limitations.

The video proves an event occurred. The evidence explains what the event means.

That difference matters because faster acquisition pathways do not make immature technology mature. In its July 2026 annual weapon-systems assessment, GAO reported that 18 of 40 programs entering the Middle Tier of Acquisition pathway between 2018 and 2025 began with immature technologies. Seven of eight current programs GAO examined still had immature technologies requiring further development.

Opening more test infrastructure can help close that gap. It will do so only if the event is designed to generate evidence rather than validate a sales narrative.

Five Things a Range-Ready Vendor Brings

The concierge can help a company navigate the range. It cannot supply the company’s test logic. Before requesting a slot, a credible vendor should be able to produce five things.

1. A Falsifiable Performance Claim

“AI-enabled,” “autonomous,” “open architecture,” and “combat proven” are descriptions. None is a test objective.

A useful claim specifies the mission task, operating conditions, measure, threshold, and failure boundary. What must the system do? Under what terrain, weather, spectrum, threat, load, or operator conditions? What result counts as success? What result triggers redesign, additional testing, or a stop?

If every possible outcome can be called promising, the event is not a test.

The claim should also match the maturity of the technology. An early prototype may be testing an interface or a navigation behavior. A production candidate may need to prove repeatability, reliability, interoperability, and maintainability. Do not ask one event to answer every lifecycle question.

2. Representative Conditions

The Army is offering access to environments commercial companies cannot easily reproduce. Use them deliberately.

Representative does not mean maximally difficult on day one. It means relevant to the claim and sequenced so the team learns. Start with a controlled baseline, add one stressor at a time, then combine conditions once the failure modes are understood.

For an autonomous or counter-UAS system, that may include degraded positioning, contested spectrum, clutter, weather, terrain masking, multiple targets, operator workload, or interoperability with another system. The exact conditions depend on the mission and the authority governing the event.

The point is not to create spectacle. It is to expose the assumptions that a clean demonstration leaves untouched.

3. An Instrumentation and Data Plan

If the team waits until the test card is written to decide what data it needs, it is already late.

The plan should identify data sources, sampling rates, time synchronization, telemetry interfaces, ground truth, environmental data, operator actions, calibration status, anomaly codes, metadata, storage, access, and post-test analysis. It should also establish who owns each data product and what can be shared with the government, partners, or future evaluators.

This is where the digital engineering ecosystem becomes practical. The digital thread is not a diagram displayed at a review. It is the trace from requirement to configuration, test condition, observation, analysis, defect, change, and retest.

Without that trace, “fix” in “fly, fail, fix, fly” becomes a guess.

4. A Configuration-Controlled Test Article

Hardware serial numbers are only the beginning.

The tested baseline should include firmware, software, model, training-data or reference-data version where applicable, sensor configuration, calibration, payload, interface definition, operator procedure, and relevant support equipment. If the team changes code or hardware during the event, record the change and identify which runs used which baseline.

This is especially important for autonomy and software-intensive systems. A result belongs to the configuration that produced it. Change the model, prompt, perception stack, data source, interface, or control logic, and some part of the evidence may need to be re-established.

A team that cannot reconstruct its tested configuration cannot credibly explain why performance improved or regressed.

5. A Transition Decision and an Owner

Every event should answer a decision for someone.

The decision might be to authorize a larger test, integrate with a government architecture, enter an acquisition pathway, fund a limited quantity, correct a specific deficiency, or stop investing. The answer may be “not yet.” That is still useful if the evidence is credible.

Name the decision owner before the test. Define the evidence that person needs. Identify the likely next authority, funding source, integration partner, and schedule if the result supports moving forward.

A range cannot manufacture a customer. It cannot create a requirement, appropriation, cyber authorization, safety release, spectrum approval, or sustainment plan simply because the system performed well.

Range access without a transition owner produces technically interesting orphan data.

The Evidence Package Is the Product

The hardware may be what enters the range. The evidence package is what leaves it.

At minimum, that package should preserve:

• the test objective and decision supported;
• the exact tested configuration;
• approved conditions and deviations;
• raw and processed data with metadata and ground truth;
• the analysis method and assumptions;
• anomalies, failures, operator interventions, and unresolved questions;
• results against the predefined measures and thresholds; and
• the recommended next decision and evidence still required.

That package does not have to be enormous. It has to be complete enough that a qualified reviewer can reproduce the reasoning and understand the limits.

This is the same principle I applied in AI at Wartime Speed: velocity and rigor are not opposites when evidence is designed into the cycle. The fastest path is usually the one that makes the next decision obvious.

A range event that produces only a video was a demonstration. A range event that changes a decision produced evidence.

The Honest Counterweights

The Army’s move is good policy. It is not magic.

First, demand can outrun capacity. Range time, airspace, spectrum, instrumentation, threat assets, safety personnel, data analysts, and qualified test staff are finite. The Army acknowledged that a request does not guarantee the desired place or time.

Second, easier entry does not remove the controls that make a government range credible. Safety, security, spectrum, airspace, environmental, foreign participation, data handling, and classification requirements will still shape what can be tested and how. International testing also adds host-nation, export, foreign-person, photography, and releasability questions. The official announcement says a company may submit without a federal relationship. It does not say every proposed event will be accepted or every capability can be exposed to every test condition.

The ATEC request-for-test-services guidance tells customers to submit early so resources can be planned and safety, security, and environmental concerns resolved. It also distinguishes ordinary range or developmental-test support from independent evaluations, operational assessments, and safety documents. If Soldiers will handle an experimental system, a separate safety release may be required.

Third, faster does not mean free. Applicants should resolve the funding and legal mechanism, direct and indirect costs, cancellation or retest exposure, and data and intellectual-property terms before committing publicly. The authorities and agreements can differ by site; companies should not assume every stateside and international location operates under one common commercial-use rule.

Fourth, opening the gates may increase pressure on an already constrained test enterprise. GAO reported in June that workforce reductions had reduced DOT&E’s capacity to oversee operational and live-fire testing across 173 programs, with remaining personnel absorbing more work and sometimes covering areas outside their expertise. The report warned of greater risk that operational shortfalls could go undocumented. That is a capacity warning, not an argument against access.

Finally, a successful range event is not government endorsement. The result belongs to the tested article, conditions, data, and analysis. Vendors should be precise about what the evidence supports and disciplined about what it does not.

The answer to these constraints is not to close the ranges. It is to schedule deliberately, standardize evidence packages, reuse instrumentation and test methods, and reserve the most demanding resources for the decisions that need them.

So What

• If you are a commercial vendor: do not begin with the portal. Begin with the decision, claim, configuration, conditions, and data you need. Then ask the concierge which range can support that test.
• If you manage a program: put a government decision owner beside the vendor before the event. Define what evidence would justify the next dollar and what result would stop or redirect the effort.
• If you operate a range: publish reusable readiness templates, interface requirements, data standards, lead times, and example evidence packages. A clear admission standard will save more time than an open calendar.
• If you lead acquisition: measure this initiative by decisions accelerated, deficiencies discovered, retest cycles shortened, and capabilities transitioned—not by portal submissions or test events completed.
• If you work in test and evaluation: protect the distinction between access and adequacy. Faster scheduling is useful. Decision-grade evidence remains the mission.

The Army has taken a meaningful step. It lowered the cost of getting a promising system into a real test environment and gave nontraditional companies a front door they did not have before.

Now comes the part no portal can automate: asking the right question, designing the right test, and preserving enough evidence to make the next decision with confidence.

The gate is open. Bring evidence.

Tags: ATECcUASDefense InnovationTest DataUAS