We need to make industry crash 100,000 drones next year

Jordanian Armed Forces service members and U.S. Army service members select a target while aiming their Dronebuster during Eager Lion 2024 in Jordan, May 13, 2024. (U.S. Army photo by Sgt. Ivan Botts)

The Pentagon is working to grow an American drone industrial base capable of producing hundreds of thousands of systems next year. Great! But what happens if we succeed?

While drones are much like munitions in the numbers the US military will need to expend, we shouldn’t stockpile drones in warehouses. Technology, threats and tactics are changing too quickly. Large inventories of today’s best drones could become tomorrow’s stockpile of obsolescence .

Instead, we need to use them. And I don’t mean just for training. I mean, really use them, run them quite literally into the ground, and then see what we can learn.

The Defense Innovation Unit’s Deputy Director, Travis Metz, told the Senate Armed Services Committee this year that military training should consume drones in “the hundreds of thousands per year.” That’s an enormous number, and it creates an enormous opportunity. We need to use the elevated demand for those drones to train the industrial base to be able to support a war that quickly expends and iterates those platforms.

The industrial base can’t just be seen as a supplier to the military. As the rate of technological change accelerates, manufacturing capacity and its ability to quickly adapt will become an increasingly integral part of the force generation system. That means industry needs to be part of the exercise.

And scale matters. We won’t know if the industrial base can adapt at scale by ordering five prototypes — we need to know whether a system can be redesigned, built, trusted, delivered and employed in operationally useful quantities. You can’t do that without producing at mass.

The good news is, we’re already on the path. I attended both of the first two Drone Dominance Gauntlets, and the improvement between them was impressive. First, the drones were better and more reliable. The operators were better too. The Marines I watched had developed checklists and common language that helped them move quickly through setup and into flight. The test team was also better. Requirements were more clearly defined, scenarios were better differentiated and data collection and scoring were substantially more sophisticated.

But most impressively, the relationship between the vendors and operators had also changed. At times they looked less like customer and supplier and more like a single tactical-technical team. The operators knew what they wanted to do while vendors were able to show them how to make the technology accomplish that. From Gauntlet 1 to Gauntlet 2, the whole drone ecosystem adapted and improved. The important lesson: By consuming drones at significant scale, we simultaneously train the force, determine the next iteration of design and create the demand to build that next iteration.

So, what comes after the four Gauntlets conclude? Hopefully, we get more Gauntlets for different classes of systems. But we better not stop there. We should also take the most important lesson and institutionalize it.

If we think the only important training is the training done by warriors on the battlefield, we are losing sight of a very important facet of modern warfare. We need to exercise the industrial base as well.

Of course, consuming drones at this scale will not be cheap. But before you accuse me of wanting to set suitcases filled with $1,000 bills on fire, we need to recognize that drones stored in a warehouse can depreciate in combat value as technology, threats and tactics change. I’m not suggesting we purchase an extra 100,000 drones just to blow them up in tests; we need to plan to crash and replace some portion of the large quantities we’re already buying.

A drone deliberately consumed in a well-designed exercise can produce a different kind of combat value from the one stored in the warehouse. We have always recognized that spent munitions and jet fuel produce higher operator proficiency, test data, better tactics and exercised logistics. But by intelligently consuming drones, we will also benefit from industrial learning and improvements to the next generation of systems. Our objective isn’t to crash as many drones as possible, nor should it be to save every one of them for war. We need to find the balance between inventory and deliberate consumption that produces the most combat value for the money.

We already have many of the pieces. The Drone Dominance Program is testing both drone performance and manufacturing readiness. Project Flytrap puts emerging technology into soldiers’ hands in operationally representative environments, and it’s already connected through a series of exercises. Major exercises provide increasingly realistic operational problems, and the Department of War is experimenting with ways to assess manufacturing readiness. Now we need to connect them and integrate manufacturing between them. Together, they could become a continuous industrial exercise that teaches how to adapt when we find ourselves in the next major conflict.

Imagine a major exercise in which thousands of drones are treated as fully consumable. Both sides have drones and counter-drone systems. Operators employ their systems while testers collect data. Engineers, supply-chain experts, acquisition professionals and manufacturers are connected to what is happening in the field.

Once one side finds a significant advantage, that may be the end of that operational event, but that would then trigger the next phase of the industrial exercise. We start the clock. Operators and testers determine what went right and what went wrong. Maybe they had an advantage in tactics but a disadvantage with the software interface. Maybe an adversary found a vulnerability in a communications link. Maybe a sensor, payload, antenna, component, or airframe needs to change.

Industry takes these lessons and immediately starts working.

Can it redesign the system? Can suppliers support the change? Can the government establish enough trust in the modified system to use it? Can the factory reconfigure? Can it produce the new version in quantities that matter? Can logistics get it back to the unit? Can operators learn the changes?

Some parts of the system will likely adapt faster than others. Finding this disparity is exactly what we need. The exercise will expose the bottlenecks that need to be fixed.

Then train again. And again. And again. Faster and faster. This could be done through a string of already existing exercises, modified to collect the right data and insert vendors at the tactical level. One exercise could discover a vulnerability, industry could adapt between events, and the next exercise could test the answer.

In effect, the industrial exercise never really ends. Individual exercises and events become iterations in a continuously adapting campaign to see what failed, what changed, at what scale, how long each step took and where we need to intervene.

Expending drones on this scale will create problems of their own. We’ll have to figure out how to recover systems, harvest useful components, handle damaged batteries, recycle materials and dispose of what cannot reasonably be recovered. Good. Those are wartime logistics problems and it’s time to start learning those lessons too.

We can’t simply replicate what Ukraine did in their drone war. The United States would likely face different constraints and would integrate drones into Joint All-Domain Operations. What we should borrow is Ukraine’s rate of learning and adapting drones to fit their operational needs.

We train pilots by flying, infantry by shooting and staffs through exercises because we don’t want to discover during wartime whether they can perform their missions. If manufacturing and adaptation are increasingly becoming part of aggregate combat power, that same logic should apply to our industrial base.

The objective shouldn’t be to enter the next war with the world’s largest warehouse of drones. It should be to enter it with the world’s most practiced system for designing, building and delivering drones. And it might take a whole lot more than crashing 100,000 drones to get there.

Col. Dustin Thomas is the Commander of DCMA’s Unmanned Systems – Experimental Command (US-X). US-X provides contract oversight for unmanned systems of all sizes across all domains and manages the Blue List. Col. Thomas previously led Black Phoenix, an innovation team focused on designing, building and flying small unmanned systems in less than 24 hours, and founded Game of Drones, an experimentation event focused on the rapid integration of new technology into unmanned systems.

The views expressed are those of the author alone and do not necessarily reflect the views of the Department of War.

2024年5月13日，在约旦举行的“渴望雄狮2024”军演中，约旦武装部队官兵和美国陆军官兵正在使用无人机拦截器瞄准目标。（美国陆军中士伊万·博茨摄）

五角大楼正致力于发展美国无人机产业基础，目标是明年生产数十万套系统。太好了！但如果我们成功了呢？

虽然无人机的数量与美军所需的弹药数量非常相似，但我们不应该在仓库里大量囤积无人机。技术、威胁和战术变化太快。今天最先进的无人机库存，明天可能就会变成过时的装备。

相反，我们需要使用它们。我指的不仅仅是训练。我的意思是，真正地使用它们，把它们用到报废，然后看看我们能从中学习到什么。

国防创新部门副主任特拉维斯·梅茨今年在参议院军事委员会作证时表示，军事训练每年将消耗“数十万架”无人机。这是一个庞大的数字，也蕴藏着巨大的机遇。我们需要利用对无人机日益增长的需求，来训练工业基础，使其能够支持一场需要快速消耗和迭代这些平台的战争。

工业基础不能仅仅被视为军队的供应商。随着技术变革速度的加快，制造业能力及其快速适应能力将日益成为军队生产体系不可或缺的一部分。这意味着工业界必须参与到这一过程中来。

规模至关重要。仅仅订购五个原型机并不能告诉我们工业基础能否大规模适应——我们需要知道一个系统能否被重新设计、制造、验证、交付并以实际运营所需的数量投入使用。而这一切都离不开大规模生产。

好消息是，我们已经走上了正轨。我参加了前两届无人机优势挑战赛，两届之间的进步令人印象深刻。首先，无人机的性能更优、更可靠。操作员的水平也更高。我观察到的海军陆战队员们制定了检查清单和通用术语，这帮助他们快速完成设置并投入飞行。测试团队也更加出色。需求定义更清晰，场景区分更明确，数据收集和评分也更加完善。

但最令人印象深刻的是，供应商和运营商之间的关系也发生了变化。他们有时看起来不像客户和供应商，而更像是一个战术技术团队。运营商清楚自己想要做什么，而供应商则能够向他们展示如何利用技术来实现这些目标。从第一届“挑战赛”到第二届“挑战赛”，整个无人机生态系统都得到了调整和改进。重要的经验是：通过大规模使用无人机，我们可以同时训练部队、确定下一代设计方向，并创造制造下一代产品的需求。

那么，四场挑战赛结束后会怎样呢？希望我们能看到更多针对不同类型系统的挑战赛。但我们最好不要就此止步。我们还应该吸取最重要的教训，并将其制度化。

如果我们认为唯一重要的训练就是战士在战场上的训练，那就忽略了现代战争中一个非常重要的方面：我们也需要锻炼工业基础。

当然，如此大规模地使用无人机成本不菲。但在你指责我想把装满千元大钞的行李箱付之一炬之前，我们需要认识到，随着技术、威胁和战术的变化，存放在仓库里的无人机的作战价值会不断贬值。我并不是建议我们额外购买10万架无人机，仅仅为了在测试中把它们炸毁；我们需要考虑的是，如何应对无人机坠毁的情况，并替换掉我们已经大量采购的部分无人机。

在精心设计的演习中故意消耗掉的无人机，其作战价值与存放在仓库中的无人机截然不同。我们一直都认识到，废弃弹药和航空燃油能够提升操作人员的熟练度，积累测试数据，改进战术并完善后勤保障。但通过合理地消耗无人机，我们还能从工业经验中获益，并助力下一代系统的改进。我们的目标并非尽可能多地击落无人机，也不应该是把每一架都留到战争中使用。我们需要在库存和合理消耗之间找到平衡点，从而最大限度地利用资源创造作战价值。

我们已经具备了许多要素。“无人机优势计划”正在测试无人机的性能和制造准备情况。“捕蝇草计划”将新兴技术应用于具有实战代表性的环境中，并通过一系列演习将两者连接起来。大型演习会提出越来越贴近实战的挑战，而美国陆军部正在探索评估制造准备情况的方法。现在，我们需要将它们连接起来，并将制造环节整合起来。这样一来，它们就能形成一个持续的工业演习，教会我们如何在下一次重大冲突中迅速适应。

想象一下这样一场大型演习：数千架无人机被视为完全消耗品。双方都拥有无人机和反无人机系统。操作人员操控系统，测试人员收集数据。工程师、供应链专家、采购人员和制造商都密切关注着现场情况。

一旦一方取得显著优势，此次作战演练可能就此结束，但同时也会触发下一阶段的工业演练。我们开始计时。操作人员和测试人员会评估哪些环节做得好，哪些环节做得不好。或许他们在战术上占优，但在软件界面方面处于劣势。或许对手发现了通信链路中的漏洞。或许需要更换传感器、有效载荷、天线、组件或机身。

行业吸取这些教训后，立即开始行动。

它能否重新设计系统？供应商能否支持变革？政府能否对改进后的系统建立足够的信任以投入使用？工厂能否重新配置生产线？它能否生产出足够数量的新版本？物流能否将其运送回单位？操作人员能否掌握这些变化？

系统中某些部分的适应速度可能会比其他部分更快。找出这种差异正是我们所需要的。这项演练将揭示出需要解决的瓶颈问题。

然后再次训练。一遍又一遍。越来越快。这可以通过一系列现有的演练来实现，只需进行修改以收集正确的数据，并在战术层面引入供应商即可。一次演练可以发现漏洞，行业可以在两次演练之间进行调整，而下一次演练则可以检验调整后的方案。

实际上，工业演练永无止境。每一次演练和事件都成为不断迭代的行动，旨在持续调整，以了解哪些环节失败了，哪些环节发生了变化，变化的规模如何，每一步耗时多久，以及我们需要在哪些方面进行干预。

如此大规模地使用无人机会带来一系列问题。我们必须想办法回收系统、拆卸可用部件、处理损坏的电池、回收材料，以及处置那些无法合理回收的物品。很好。这些都是战时后勤问题，我们也应该开始吸取这些经验教训了。

我们不能简单地照搬乌克兰在无人机战争中的做法。美国可能会面临不同的限制，并将无人机整合到全域联合作战中。我们应该借鉴的是乌克兰学习和调整无人机以适应自身作战需求的速度。

我们通过飞行训练飞行员，通过射击训练步兵，通过演习训练参谋人员，因为我们不想在战时才发现他们是否能够胜任任务。如果制造和适应能力日益成为整体作战能力的一部分，那么同样的逻辑也应该适用于我们的工业基础。

目标不应该是带着世界上最大的无人机仓库投入下一场战争，而应该是带着世界上最成熟的无人机设计、制造和交付系统投入战争。而要实现这个目标，恐怕需要的远不止是击落10万架无人机。

达斯汀·托马斯上校是美国国防合同管理署（DCMA）无人系统实验司令部（US-X）的指挥官。US-X负责监管所有领域各种规模的无人系统合同，并管理蓝名单。托马斯上校曾领导“黑凤凰”创新团队，该团队致力于在24小时内设计、建造和试飞小型无人系统；他还创立了“无人机游戏”（Game of Drones）实验活动，该活动旨在快速将新技术集成到无人系统中。

文中表达的观点仅代表作者个人观点，并不一定反映美国战争部的观点。