Imagine you run a small AI startup. You have a tool that can process satellite imagery, predict supply chain failures, or help a commander decide what to strike. The Pentagon might want it, but the idea of selling to the government sounds like a nightmare: endless paperwork, audits, compliance reviews, and a procurement process that can take longer than your company has existed. Now imagine instead that you just have to make a five-minute video. No 200-page proposal, no mandatory security clearance, no waiting for a request for proposals. You film yourself explaining what your product does, upload it, and wait for a panel of judges at the Department of Defense to say yes or no. That’s the reality of Tradewinds, an under-the-radar procurement program run by the DOD’s Chief Digital and Artificial Intelligence Office, or CDAO. Since late 2022, Tradewinds has been quietly building a catalog of AI products that government buyers can purchase quickly. Companies submit videos addressing “strategic focus areas” that the Pentagon updates periodically. At least once a month, judges review the submissions and decide whether to accept them into the Tradewinds Solutions Marketplace, a kind of curated shopping catalog for government officials. If accepted, the product is labeled “post-competitive,” a special status that lets government buyers skip many of the competitive bidding requirements that normally slow everything down. The effect can be dramatic. A DOD official says that in several cases, awards have been made in less than a week. For a system where “fast” often means a year, that’s almost unbelievable. The marketplace includes some of the biggest names in AI—OpenAI, Anthropic, and Google—though all three declined to comment for this story. It’s a strange new world: the most powerful military in history, choosing its next AI tools based on YouTube-style pitches.
Tradewinds didn’t appear out of nowhere. It’s a response to a crisis the Pentagon helped create. For decades, the US military funded research and development that gave birth to the modern tech industry, but it also allowed the defense supplier base to shrink into a handful of giant prime contractors. Small companies found it nearly impossible to break in. The barriers weren’t just technical; they were bureaucratic. Compliance costs, security requirements, and a risk-averse culture made the Pentagon a terrible customer for agile startups. If you’re a founder with a working AI product, why would you spend 18 months chasing a contract that might never come, when you could sell to a commercial client in weeks? The military knows it has a problem, and Tradewinds is its attempt to inject competition into a market it essentially destroyed. The phrase “desperate bid” is apt. The Pentagon isn’t just opening a door; it’s tearing down walls. The Trump administration has been far more vocal than its predecessors about buying and using AI, and it wants the budget to match. In its 2027 budget request, the White House asked Congress for $1.5 trillion for the Department of Defense, citing “historic investments” in AI as a priority. That’s a massive amount of money, and Tradewinds is one of the main channels the Pentagon wants to use to spend it. But here’s the catch: the same features that make Tradewinds attractive—speed, flexibility, fewer rules—also make it harder to track. When you move fast and break the old procurement norms, you also leave less of a paper trail. The Pentagon is betting that the benefits of agility outweigh the risks of opacity. It’s a bet that could define the future of military technology.
The most striking thing about Tradewinds isn’t just how it works; it’s what the Pentagon is asking for. Under the Biden administration, the final Tradewinds submissions cycle focused on products that promoted “a culture of responsible AI” and provided guardrails for AI development and use. That was the language of caution, ethics, and risk management. This year, the military is asking for something very different. The 2026 strategic focus areas include AI that can “enhance force lethality” and “unleash” AI-enabled agents that can assist in the joint targeting process—including so-called kill chain execution. In plain English, the Pentagon wants AI that can help find targets, decide what to strike, and execute attacks faster than a human team could. A DOD official says the 2026 topics were chosen to align with a January 2026 AI strategy memo, but the shift in tone is impossible to ignore. “Responsible AI” hasn’t disappeared entirely, but it’s no longer the centerpiece. The message to startups is blunt: if you want to sell to the military, don’t lead with ethics, lead with lethality. This is a profound cultural shift, and it raises uncomfortable questions. Are guardrails still part of the conversation, or are they just overhead? Can an AI system that autonomously helps execute a kill chain ever be considered “responsible”? The Pentagon would likely say yes, because humans are still in the loop, at least for now. But the emphasis on speed and kill chains suggests that the priority is capability over caution. For startups watching these signals, the path is clear: build tools that make the military more lethal, and you’ll find a welcome mat. Build tools that make AI safer, and you might be waiting a long time for a call back.
The excitement around defense tech is not confined to the Pentagon. Daniel Driscoll, then the sitting US Army secretary, wrote an unusual pitch to startups as part of Y Combinator’s Fall 2026 wishlist of companies to fund. “There has never been a better time to build for the Army,” he wrote, urging entrepreneurs to see the military as a customer. Driscoll has since resigned and returned to the private sector, and he didn’t respond to a request for comment, but his words capture the mood of a growing movement. Venture capitalists and founders are starting to see defense as one of the hottest markets in tech, and programs like Tradewinds are a big reason why. The “post-competitive” status and the use of “other transaction” agreements, or OTs, make it easier for the government to buy from small companies without the usual federal acquisition rules. OTs are fast, flexible, and not subject to many of the requirements that make traditional contracts so slow. For a startup, that’s a huge relief. You don’t have to gamble on whether a months-long procurement process will actually result in a contract. You can get a decision in weeks, maybe even days. But there’s a downside. Not all agencies publicly report their OT spending, and even when they do, the data is hard to find on government tracking sites like USASpending.gov and SAM.gov. The Pentagon can legally spend up to $500 million on a single OT without notifying Congress. That’s an enormous amount of money moving through a system that is deliberately opaque. For startups, that opacity is a feature: less paperwork, faster checks, fewer questions. For taxpayers and watchdogs, it’s a bug: no clear way to know what the military is buying, why, and whether it’s worth the price. The tension between speed and accountability is at the heart of Tradewinds.
What does all this mean for the future of military AI? On one hand, Tradewinds is a genuinely clever response to a real problem. The old procurement system was designed for tanks, jets, and ships—not for software that changes every few months. By the time a traditional contract was awarded, the AI technology on offer might be outdated, or the startup might have gone out of business. Tradewinds creates a marketplace where the government can see what’s available, compare options, and buy quickly. That’s a huge improvement for anyone who believes the military should have access to cutting-edge technology. It also gives small companies a chance to compete with the usual prime contractors, which could bring fresh ideas and lower prices. On the other hand, the program’s success depends on the judgment of a small panel of judges and the discretion of procurement officers. There’s no public scorecard showing which products are accepted and why, no clear criteria for what counts as “post-competitive,” and no easy way to audit the deals that result. The same speed that makes Tradewinds attractive also makes it vulnerable to hype and groupthink. A five-minute video can be polished and persuasive, but it’s not the same as a rigorous technical evaluation. A founder can say anything in a pitch; the government has to live with the consequences. The Pentagon’s willingness to embrace AI is understandable. Adversaries are developing their own autonomous systems, and the US doesn’t want to fall behind. But the shift from “responsible AI” to “force lethality” is not just a bureaucratic change. It’s a statement of values. The military is signaling that it wants AI that can kill, and it wants it now. Whether that’s a good idea is a question that deserves more than a five-minute video.
In the end, Tradewinds is a story about speed, money, and power. The Pentagon is trying to buy its way out of a procurement crisis by creating a parallel system that bypasses the old rules. It rewards startups for being concise, persuasive, and aligned with the military’s current priorities. The program has already attracted major players like OpenAI, Anthropic, and Google, and it’s likely to grow under a Trump administration that wants to spend $1.5 trillion on defense and make AI a centerpiece of that spending. But the cost of that speed is transparency. Other transaction agreements can be used to spend up to $500 million without congressional notification, and the public records of these deals are spotty at best. That’s a dangerous combination: a lot of money, a fast process, and little oversight. It’s also a reflection of a broader trend in American defense policy, where the urgency of great-power competition is used to justify cutting corners on accountability. Tradewinds may be a clever tool, but it’s not a solution to the underlying problem. The US military destroyed its own supplier market through decades of consolidation and bureaucracy, and a video-pitch program can’t rebuild trust overnight. What it can do is open the door to a new generation of startups—and to a new era of military AI that is faster, more lethal, and much harder to question. Whether that’s a future we want is a decision that shouldn’t be left to a monthly panel of judges in a quiet Pentagon office. It’s a decision for all of us, and right now, we’re not being asked. The five-minute videos are rolling, the money is moving, and the questions are only beginning.
