The Pentagon is quietly overhauling one of its most entrenched systems: how it sources and buys modern technology — and by extension, who controls it.
After a public split with Anthropic, the Defense Department is cutting deals across the tech industry, assembling a line-up of top-performing commercial players to deploy their models on its classified networks.
In this emerging model, different layers of capability come from different firms: connectivity and battlefield resilience are expected from SpaceX; frontier model intelligence from OpenAI and Google; compute and hardware acceleration from Nvidia; open-weight model ecosystems from Reflection; and enterprise deployment at scale from Microsoft, Oracle and AWS.
The Pentagon is billing this pivot toward a “best of breed” strategy – assembling a mix of commercial providers instead of betting on a single vendor – as a way to accelerate its “artificial intelligence-first” warfighting vision.
But read more closely, it’s something bigger: a deliberate move to avoid winner-take-all dominance and retain long-term control over the technologies that will define modern warfare.
From defense giants to modular power
For decades, Pentagon contracts were dominated by a small group of entrenched defense primes. That system delivered reliability at scale, but it also came with familiar downsides: cost overruns, slow innovation and limited competition. Once a vendor was chosen, the government was effectively locked in.
What’s changing now is not just who builds military systems, but how they’re built.
Instead of monolithic, all-in-one platforms, the Pentagon is shifting toward a model that looks more like modern cloud architecture: modular, interoperable and composed of multiple vendors. No single company owns the full stack, and that’s the point.
By distributing capability, the Pentagon reduces its dependence on any one provider, especially those offering proprietary systems it can’t fully inspect, audit or replace. The tradeoff is real: integration becomes harder, and coordination more complex. But the alternative, opaque systems controlled by a single vendor, creates a different kind of risk that is much harder to unwind.
Control isn’t just strategic – it’s evolutionary
As AI becomes embedded in military decision-making, the question of control becomes more than a technical issue. It’s a governance issue. Military AI systems will shape life-and-death decisions, and in a rule of law system, they can’t be black boxes. They need to be subject to meaningful scrutiny by government operators, auditors and ultimately elected officials.
That’s where the Pentagon’s multi-vendor approach starts to look less like procurement reform and more like a safeguard. Overreliance on closed, proprietary models risks concentrating power in private hands, limiting visibility into how decisions are made. A modular system, by contrast, creates conditions for accountability.
Inspectable, open-weight models are especially important in that equation. They allow the government to test for bias, enforce rules of engagement, and ensure systems align with United States law and international norms. Closed systems, however advanced, introduce a fundamental tension: The military may deploy tools it doesn’t fully understand.
Why open-weight models matter
There is also a geopolitical dimension to this shift.
Companies like DeepSeek, Alibaba, Tencent, Zhipu AI, and Huawei-linked infrastructure are being folded into China’s military ecosystem, supporting everything from surveillance and target recognition to war-gaming and command-and-control functions. Chinese-developed open models are already spreading across global software ecosystems – and in many cases, sovereign customers and developers are adopting them simply because they’re affordable, accessible and customizable.
That creates a subtle but powerful form of influence. If U.S. institutions rely primarily on closed systems while open ecosystems elsewhere continue to expand, the balance of technological gravity could shift.
The Pentagon appears to recognize this risk. Supporting open-weight alternatives isn’t just about flexibility; it’s about ensuring that the global AI ecosystem isn’t tilted toward Beijing.
A quiet but consequential shift
In that sense, the Pentagon’s new AI multi-partner strategy is about far more than modernization. It’s a structural shift away from concentrated power – both the old defense primes and the risk of a new generation of tech monopolies – toward a system that preserves competition, visibility and control.
The ability to swap vendors, benchmark performance and avoid lock-in gives the Pentagon something it hasn’t consistently had in decades: negotiating power and flexibility as technology evolves. More importantly, it creates the foundation for transparency and auditability to exist in practice, not just in policy.
That’s the deeper story behind this shift. It’s not just about adopting AI faster. It’s about preventing any single company – old traditional defense giants or new tech leaders – from becoming indispensable.
For a military increasingly defined by software, that’s not just a strategic advantage.
It’s a democratic one.
Roslyn Layton, PhD, senior vice president of Strand Consult and visiting researcher at Aalborg University Copenhagen, is an international technology expert focused on its intersection with economics, security and geopolitics. She is a fellow of the National Security Institute at George Mason University and a senior advisor to the Foundation for American Innovation and has testified before the U.S. Congress on multiple occasions.
Copyright
© 2026 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.
