Defense Business Brief: Tech at the edge; PACFLEET talks cyber; and more

Defense Business Brief

HONOLULU— Soldiers in the Pacific need technology that works when networks are attacked and connectivity isn’t guaranteed, said Gen. Xavier Brunson, the commander of U.S. Forces Korea.

“If the service tells you who they are and what they're going to do, and you choose not to try to meet the requirements of how they're going to fight—that's going to be a problem. Because in the future, our networks will be fully under attack. I think in the future we'll have to operate disaggregated and distributed. I think in the future, things will happen so fast, they'll happen so fast that we won't know sometimes what's happening to us,” Brunson warned during a keynote Tuesday at the TechNet Indo-Pacific conference here.

“There will be plans that will change before we even get in the theater…And if your product can't operate on the edge, then it's not a product that's going to be able to help the Army fight the way the Army's going to fight, much less the joint force.”

Brunson’s comments come as what’s known as the Davidson window , a 2027 timeline for when China’s leader has told its military it must be ready to invade Taiwan, draws near.

But while there’s no expectation of a war with China next year, it’s something the military is still preparing for, Gen. Ronald Clark, commander of U.S. Army Pacific Command, said in his keynote address.

“There are 157 days between today and 2 April. So that mark on the wall for us in our headquarters is to ensure that we are in position, postured forward to deter war in a window in which our adversary may find it to their advantage to attack Taiwan. But again, it's not a date that we think that conflict will happen…But we'll continue to prepare. We'll continue to modernize. We'll continue to posture ourselves for the business of deterred war well beyond [2027].”

Congress and the lack of a 2027 budget could stymie some of those modernization efforts unless lawmakers pass a full budget by December.

Working under stopgap funding is the “biggest challenge” that affects “plans for new starts and being able to fund some of the programs that we know are going to yield benefit to the soldier,” Clark told reporters earlier this month.

Examples include experimentation and some efforts to improve the mission partner network , he said: “You can't start that contract if you don't have new money. So it's a challenge that, again, we have to overcome…we have more experience operating under a [continuing resolution] than we do with an appropriation.”

You’ve reached the Defense Business Brief, where we dig into what the Pentagon buys, who they’re buying from, and why. Send along your tips, feedback, and mai tai recommendations to lwilliams@defenseone.com . I live here now. Check out the Defense Business Brief archive here , and tell your friends to subscribe !

PACFLEET’s cyber needs. The Navy needs more continuous monitoring solutions for systems on ships and ashore, including products that scan for vulnerabilities, Richard Williams , who manages command, control, communication, computers, and intelligence platforms for the U.S. Pacific Fleet, said Tuesday.

“I would offer that we are in a zero day today. We do need more continuous monitoring, both at the shore and our ship nodes,” Williams said during a panel discussion. “We struggle with scanning and patching vulnerabilities within our networks, and having that partnership with industry and academia on how we can move quicker to do that, I think will really help us.”

Adding to that complexity is the need to tailor cybersecurity to different ships: Destroyers and aircraft carriers “have very various mission sets. So we need to be cognizant of what those missions are and how we can support them with cybersecurity,” he said. “So I think from industry and enterprise level, with regards to cyber, helping us with supporting us with technology to do continuous monitoring, patches and scanning, and allowing us to move quicker to keep pace with our adversaries.”

Allies + AI. Each nation has a different approach to large language models, or LLMs, which adds another layer of complexity to the work required for the U.S. military to be on the same page as allies and partners.

“We've got different approaches to LLMs in our respective countries,” including planning with a different ecosystem and decision-making process, said Australian army Brig. Chris Alder, the vice director for intelligence at Indo-Pacific Command.

The first challenge is engaging with partners, he said during a panel Wednesday: “Understanding how [an ally’s] respective partners are approaching their own challenges is a key start point. So working through our liaisons that we've got within PACOM to get the right points of contact to have those discussions has been really important. And to be frank, knowing which industry partnerships countries are pursuing has often been accommodating for discussions.”

A little more: Nations in the Pacific have varying levels of AI adoption and implementation—and many are not using frontier models, said Scott Brunstetter, U.S. Indo-Pacific Command’s chief AI officer.

“There are some exceptions to that,” where some are building “their own smaller capabilities. So each journey has been a little bit different…it comes back to starting with foundations and how do we understand everything from governance and workforce training, things like that,” Brunstetter said.

He requested that companies proactively help bridge the divide to help the U.S. understand where partner nations are, technologically: “You all have some very interesting and insightful tools, and as you engage with partners out there, circle back with us. Help us understand where that partner was. Because the other side of this is partners don't have a whole lot of money in some cases, and so they're looking for…cheaper options. So there's a wide open opportunity there.”

A related, haunting thought: When it comes to AI, there’s a lingering concern around data security . And there’s one question senior leaders should be asking themselves more: “How would they know if the models and AI systems that they are using have been poisoned ? And it's giving them subtle and incorrect answers over time—how would you detect that?” Brad Christ , the University of Hawai’i’s chief information officer, said in a separate panel. “And I don't think that there's an answer.”

Making moves + other news

The Army expands testing sites for commercial drone use.

SAIC has a quantum portfolio.

General Atomics completed flight testing with Saab’s LoyalEye, an advanced early warning system, on the MQ-9B Reaper drone.

Energetics and propulsion startup Firehawk is now valued at more than $1.4 billion after closing a Series D funding round of $150 million. The round was led by Advent and Rochefort Asset Management.

Beacon AI’s autonomous pilot tech took flight with KC-46 tankers.

NEXT STORY: The US space industry will suffer for Ukraine’s Starlink outage, experts say

国防商业简报

檀香山——驻韩美军司令泽维尔·布伦森将军表示，太平洋地区的士兵需要一种在网络遭受攻击且连接无法得到保证时也能正常工作的技术。

“如果攻击者告诉你他们是谁，要做什么，而你却选择不去满足他们提出的应对要求——那就会出问题。因为未来，我们的网络将面临全面攻击。我认为未来我们必须采用解耦和分布式运营模式。我认为未来，事态发展如此之快，以至于我们有时根本无法察觉发生了什么，”布伦森周二在TechNet印太会议上发表主题演讲时警告说。

“在我们进入战场之前，计划就会有所改变……如果你的产品不能在极端环境下运行，那么它就无法帮助陆军以陆军的方式作战，更不用说联合部队了。”

布伦森发表这番言论之际，正值所谓的“戴维森窗口”临近，该窗口指的是2027年中国领导人指示军方做好入侵台湾的准备。

美国太平洋陆军司令部司令罗纳德·克拉克将军在主旨演讲中表示，虽然预计明年不会与中国发生战争，但军方仍在为此做准备。

“从今天到4月2日还有157天。因此，我们总部墙上的这个标记是为了确保我们做好准备，在对手可能趁机攻击台湾的窗口期，能够有效威慑战争。但我们再次强调，我们认为冲突不会在4月2日爆发……不过，我们将继续做好准备，继续推进现代化建设，继续保持威慑战争的能力，直至2027年以后。”

除非国会议员在 12 月之前通过完整的预算，否则国会和 2027 年预算的缺失可能会阻碍一些现代化努力。

克拉克本月早些时候告诉记者，在临时拨款的情况下开展工作是“最大的挑战”，这影响到“新项目的启动计划，以及为一些我们知道将对士兵有益的项目提供资金”。

他举例说，例如进行试验和努力改进任务伙伴网络：“如果没有新的资金，就无法启动合同。所以，这又是我们必须克服的挑战……我们在[持续决议]下运作的经验比在拨款下运作的经验更丰富。”

欢迎来到《国防商业简报》，在这里我们将深入探讨五角大楼的采购内容、供应商以及采购原因。欢迎将您的建议、反馈和迈泰鸡尾酒推荐发送至 lwilliams@defenseone.com。我现在就住在这里。点击此处查看《国防商业简报》往期内容，并邀请您的朋友订阅！

美国太平洋舰队的网络安全需求。负责管理美国太平洋舰队指挥、控制、通信、计算机和情报平台的理查德·威廉姆斯周二表示，海军需要更多针对舰载和岸上系统的持续监控解决方案，包括用于扫描漏洞的产品。

威廉姆斯在小组讨论中表示：“我认为我们目前正处于零日漏洞危机时期。我们需要加强持续监控，无论是在岸上还是在舰载节点上。我们在扫描和修复网络漏洞方面面临诸多挑战，而与业界和学术界合作，探讨如何更快地完成这项工作，我认为将对我们大有裨益。”

雪上加霜的是，还需要针对不同舰艇定制网络安全方案：驱逐舰和航空母舰“任务类型截然不同。因此，我们需要了解这些任务是什么，以及如何通过网络安全来支持它们，”他说道。“所以我认为，从行业和企业层面来看，在网络安全方面，我们需要借助技术来支持我们进行持续监控、补丁更新和扫描，从而使我们能够更快地行动，与对手保持同步。”

盟友 + 人工智能。每个国家对大型语言模型（LLM）都有不同的方法，这使得美国军方与盟友和伙伴保持一致的工作变得更加复杂。

“我们各自国家对终身学习采取了不同的方法，”澳大利亚陆军准将、印太司令部情报副主任克里斯·奥尔德表示，包括采用不同的生态系统和决策流程进行规划。

他在周三的一个小组讨论会上表示，首要挑战是与合作伙伴开展合作：“了解盟友各自的合作伙伴如何应对自身挑战是关键的起点。因此，通过我们在太平洋司令部内部的联络人，找到合适的联系人进行讨论至关重要。坦白地说，了解各国正在寻求哪些行业合作伙伴关系，通常对讨论大有裨益。”

美国印太司令部首席人工智能官斯科特·布伦斯特特表示，太平洋各国对人工智能的采用和实施水平各不相同，而且许多国家没有使用前沿模型。

“也有一些例外情况，”布伦斯特特说，“有些机构正在构建他们自己的小型能力。所以每个人的发展历程都略有不同……归根结底，还是要从基础做起，了解从治理到员工培训等方方面面的问题。”

他请求各公司积极主动地帮助弥合技术差距，使美国能够了解伙伴国家的技术水平：“你们都拥有一些非常有趣且富有洞察力的工具，在与合作伙伴互动时，请及时反馈给我们，帮助我们了解合作伙伴的现状。因为另一方面，有些合作伙伴资金有限，所以他们正在寻找……更经济的选择。因此，这里存在着巨大的机遇。”

还有一个与之相关且令人不安的问题：在人工智能领域，数据安全始终是一个令人担忧的问题。高层领导应该更多地扪心自问：“他们如何才能知道他们使用的模型和人工智能系统是否已被篡改？这些系统是否会随着时间的推移给出微妙且错误的答案——他们该如何检测出来？”夏威夷大学首席信息官布拉德·克里斯特在另一个小组讨论中表示：“我认为这个问题没有答案。”

最新动态 + 其他新闻

陆军扩大了商用无人机测试场地。

SAIC拥有量子技术产品组合。

通用原子公司已完成在 MQ-9B“死神”无人机上使用萨博公司的 LoyalEye（一种先进的预警系统）的飞行测试。

能源和推进系统初创公司Firehawk在完成1.5亿美元的D轮融资后，估值超过14亿美元。本轮融资由Advent和Rochefort Asset Management领投。

Beacon AI 的自主飞行技术已应用于 KC-46 加油机。

下一篇报道：专家称乌克兰星链服务中断将对美国航天工业造成影响