AI News AI资讯 8h ago Updated 1h ago 更新于 1小时前 52

Defense tech is flooded with money, but who’s built to last? 国防科技资金泛滥,但谁是长久之计?

Anduril just hit a $14 billion valuation. Mach Industries, which most of the public hasn’t heard of, is now worth $3 billion. This isn’t a bubble; it’s a declaration of a new industrial order, one where Silicon Valley’s playbook of moonshots and software margins is being violently rewritten for the messy, lethal reality of state power. The U.S. government, in a rare moment of bipartisan clarity, is signaling a 40% budget increase for defense. A gold rush is underway, and the prospectors are floo 又一场估值狂欢在国防科技赛道上演。Anduril和Mach Industries的估值像坐了火箭,一个翻倍,一个四倍。背景音是美国政府拟将国防预算再砍上40%的一刀。于是,一群新的创业公司,像嗅到血腥味的鲨鱼,开始扑向那些政府合同。但写下了Anduril第一张支票的风险投资人 Ross Fubini 直接泼了盆冷水:大多数会死在“原型合同”到“采购合同”之间的那个“死亡之谷”里。

75
Hot 热度
80
Quality 质量
70
Impact 影响力

Analysis 深度分析

Anduril just hit a $14 billion valuation. Mach Industries, which most of the public hasn’t heard of, is now worth $3 billion. This isn’t a bubble; it’s a declaration of a new industrial order, one where Silicon Valley’s playbook of moonshots and software margins is being violently rewritten for the messy, lethal reality of state power. The U.S. government, in a rare moment of bipartisan clarity, is signaling a 40% budget increase for defense. A gold rush is underway, and the prospectors are flooding in with pitch decks about autonomous systems and AI-driven targeting.

But here’s the unvarnished truth that the press releases and funding announcements obscure: this isn’t a technology race. It’s a systems race, and most of these new entrants are already dead. They just don’t know it yet. Ross Fubini, the VC who placed the first bet on Anduril, put it plainly when he warned about the “Valley of Death.” That’s not just a cute industry phrase. It’s a mass grave for startups.

The core delusion is believing that building a superior prototype—the thing Valley founders are brilliant at—is even 20% of the battle. The real war is waged in windowless Pentagon offices, in the grinding gears of the Federal Acquisition Regulation (FAR), and in the slow, paranoid digestion of institutional risk. You don’t win a $10 billion JADC2 contract with a slick demo. You win it by embedding your engineers into the Air Force’s development cycle for three years, by learning to speak the arcane language of requirements documents, and by accepting that your elegant, agile software will be bolted onto a decade-old legacy platform because that’s what the budget line item covers.

Anduril and Mach didn’t succeed just because they had good tech. They succeeded because they had founders who understood that the product wasn’t just hardware or code; the product was trust with a very specific, very risk-averse customer. They mastered the art of “dual-use” not as a marketing buzzword, but as a survival strategy—proving value to the DoD while maintaining a commercial foothold that keeps the lights on and the tech moving. They built cultures that could tolerate the glacial pace of government alongside the frantic sprint of startup iteration. That’s a cultural mutation most pure-play defense startups are completely unequipped for.

So what will happen to the new wave? Most will fizzle out in the funding gap between their Series A and their first real program-of-record contract. They’ll burn through cash trying to “sell” to the government, not realizing that the government doesn’t buy things; it awards contracts through processes that reward past performance and risk mitigation above all else. The few who make it through will do so not by being the most technically innovative, but by being the most operationally and politically adept. They’ll become, in essence, defense contractors—a role that Valley types often disdain.

This isn’t a criticism; it’s a diagnosis. The real disruption isn’t coming from a new AI model that can identify a tank faster. It’s coming from companies like Anduril that are fundamentally redesigning the business of being a defense prime. They’re applying software business models—subscription services for autonomous drone fleets, for instance—to hardware, creating recurring revenue streams that Wall Street loves and the Pentagon is slowly learning to tolerate. They’re creating a new category, one that sits uncomfortably between Lockheed Martin and a Silicon Valley SaaS startup.

The exciting frontier isn’t just more startups. It’s the pressure this places on the entire ecosystem. Can the DoD’s procurement office actually adapt to buy software like a service? Can the big primes, drowning in their own bureaucratic inertia, acquire and integrate these fast-moving entities without killing them? The real competition isn’t between Anduril and the next upstart. It’s between Anduril’s model of rapid, software-defined iteration and the decades-old defense acquisition model that has defined the industry since the Cold War.

I’m bullish on the mission. A modernized, more agile defense capability is a genuine national security necessity. But I’m deeply skeptical of the crowd chasing the money. The valley they need to cross isn’t filled with prototypes; it’s filled with paperwork, legacy systems, and cultural friction. The next great defense giant will look less like a startup that got lucky and more like a strange, new hybrid organism—part software company, part government integrator, part political entity. The frenzy is understandable, but the graveyard is filling up. The winners will be those who realize this isn’t a software problem to be solved, but a political and institutional labyrinth to be survived.

又一场估值狂欢在国防科技赛道上演。Anduril和Mach Industries的估值像坐了火箭,一个翻倍,一个四倍。背景音是美国政府拟将国防预算再砍上40%的一刀。于是,一群新的创业公司,像嗅到血腥味的鲨鱼,开始扑向那些政府合同。但写下了Anduril第一张支票的风险投资人 Ross Fubini 直接泼了盆冷水:大多数会死在“原型合同”到“采购合同”之间的那个“死亡之谷”里。

这画面,熟悉得令人不安。十年前,硅谷的热钱追逐O2O、共享单车、区块链,逻辑如出一辙:一个宏大的叙事(国防现代化),一笔确定性的巨款(政府预算),催生一批试图用技术改变“游戏规则”的玩家。然后呢?大潮退去,沙滩上留下一堆无法商业化、无法规模化、甚至无法通过军工体系冗长且苛刻验收流程的“技术尸体”。今天国防科技的热闹,多少有点“虚火”往上冒的意思。

估值翻倍,首先得问一句:凭什么?Anduril的硬件产品、自主系统和Lattice平台的确有其价值,它成功地将硅谷的敏捷开发和软件思维,注入了向来封闭迟缓的国防工业。但四倍估值的Mach Industries呢?它的核心是利用高速飞行器进行“点对点配送”——送快递?还是送炸药?故事很性感,但离成为下一个“国防领域的亚马逊”还差着十万八千里。这里面有多少是对技术的真正信仰,又有多少是对“国防预算”这块铁板上凿下来的碎屑的贪婪追逐?恐怕后者占比不小。

政府增加40%的预算,听起来是泼天的富贵。但军方的采购流程,是出了名的官僚主义迷宫,外加“非升即走”的政治正确。一个初创公司,从拿到一份小额“其他交易授权”合同,到真正在五角大楼的预算科目里,为自己争取到一份稳定、批量、利润丰厚的采购订单,中间隔着的不是山海,是官僚的深渊、是现有军工巨头的围剿、是国会山听证会上“为何要买一家不成熟小公司的产品”的灵魂拷问。Fubini说的“死亡之谷”,本质就是这个:你可以用风投的钱造出惊艳的原型,但用纳税人的钱买单,需要的是另一套截然不同的、关于可靠性、供应链、政治游说和耐心的学问。

于是,我们看到了一幅荒诞的图景:大量创业公司冲进这条赛道,不是为了去真正啃国防科技那些“硬骨头”(比如可靠降本的自主潜航器,或者能实战部署的群体智能),而是为了制造一份对政府合同有“解释力”的财报,好在下一轮融资中套现离场。他们追逐的是“国防科技”这个标签带来的估值溢价,而不是真正的国防能力。这本质上是一场以国家安全为赌注的金融游戏。真正的创新需要十年磨一剑的沉寂,而资本要求的是季度报告的增长。

更深的忧虑在于,这种狂热可能正在扭曲真正的技术创新方向。当所有聪明人都去追逐看得见的国防预算时,那些不那么“性感”但关乎未来的民用科技基础研究,谁来做?我们是不是正在用“军工复合体”的旧模式,给“风投复合体”穿上新马甲?国防科技需要的是能打仗、打胜仗的“好钢”,但现在涌入的,可能更多的是想快速回炉变现的“铁渣”。

那张通往五角大楼的船票,票价已经贵得离谱。大多数新玩家不仅买不起,连码头都找不到。热闹是他们的,大多数人什么也留不下。

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

融资 融资 政策 政策 自动驾驶 自动驾驶
Share: 分享到: