If you'd read the assigned brief for this issue, you'd expect a story about Palantir tightening its grip on Pentagon AI work while smaller rivals scramble for scraps. That's not what the record shows this week. What the Army's latest Next Generation Command and Control announcement actually reveals is something more interesting for anyone betting on defense tech startups: a program architecture where Palantir is one name among many, and a decidedly non-dominant one at that.
Nine Companies, Not One Winner
On October 7, the Army awarded roughly $93.6 million in application contracts to nine companies for NGC2 — the service's effort to build command-and-control software that scales to the corps level, starting with I Corps in the Pacific (Breaking Defense). The roster includes General Dynamics Mission Systems alongside a cluster of smaller, less household names: Air Space Intelligence Federal, Immersive Wisdom, LMI Consulting, Mente Systems, Stilman Advanced Strategies, Onebrief, Rune Technologies, and AIR, the company formerly known as Govini (Breaking Defense).
Palantir isn't on that list. It shows up earlier in the piece, in a different and more structurally revealing role: back in June, the Army tapped Anduril to lead the common data layer baseline for NGC2, with Palantir and Raft brought on as support (Breaking Defense). That's a subordinate position in the program's org chart, not a captured one. Joseph Welch, the Army's portfolio acquisition executive for C2 and Counter C2, described the application awards as "a foundation that can be rapidly scaled and trained with continuity across the force" (Breaking Defense) — language that describes a layered system built by a dozen-plus vendors, not a single-prime AI stack.
If you're pricing Palantir's moat on the assumption that it's becoming the default AI layer across Pentagon programs, this contract is the counter-evidence. NGC2 is deliberately plural. The data layer, the application layer, and the eventual integration work are being split across companies with very different pedigrees — from a 100-year-old prime in General Dynamics to startups most investors outside this beat have never heard of.
The Army's Rhetoric Is Catching Up to the Procurement Structure
The timing lines up with something else happening at AUSA this week. Acting Army Secretary Adam Telle used his first public address not to talk up any single AI vendor but to attack the vocabulary around AI modernization itself. "We've drowned ourselves and the public in buzzwords, strategies, slideshows, government acronyms and bureaucratic doublespeak," Telle told the AUSA Annual Meeting, framing the Army's acquisition reform around consolidating "fiefdoms" into a small number of portfolio executives empowered to make trade-offs on speed and manufacturability (Defense News). Telle also pointed to the Army's small-drone marketplace — the Amazon-style procurement tool — as the model he wants replicated (Defense News). That's a vote for distributed, multi-vendor procurement mechanics, not for consolidating critical software around one or two incumbents, legacy or otherwise.
Even the AI Hype Is Getting Walked Back
There's a third data point worth putting next to this, because it undercuts the premise that Pentagon AI adoption is accelerating in a straight line toward any single winner. The Defense Counterintelligence and Security Agency told Defense News this month that its "Modernizing Polygraph" effort does not use generative AI, large language models, or facial and vocal emotion analysis — walking back language from earlier 2027 budget documents and internal slides that had described exactly those capabilities (Defense News). Critics had warned the tools lacked scientific grounding and risked compounding errors in a domain with few civil-liberties protections (Defense News). That's not evidence of an AI land grab. It's evidence of an institution still negotiating, program by program, how much AI it actually wants.
Put these three threads together and the picture for smaller AI-focused defense companies isn't one of being squeezed out by a dominant prime. It's one of fragmented, multi-vendor contracting where a startup like Rune Technologies or Onebrief can win a seven-figure application slot on the same announcement where Palantir shows up only as a support player on someone else's baseline. The companies worth watching next aren't the ones trying to out-scale Palantir — they're the ones proving they can hold their lane inside a program the Army is explicitly designing not to have one winner.
