The companies—OpenAI, Google, Microsoft, Amazon Web Services, NVIDIA, SpaceX, and emerging startup Reflection AI—will now see their technologies deployed across classified, secret, and top-secret Pentagon networks. Several of these firms already maintain longstanding relationships with the military, but this expansion deepens their reach into high-security operations.
Notably absent from the list is Anthropic, a major AI developer that has recently fallen out with the Pentagon. Earlier this year, the Defense Department labeled Anthropic a “supply-chain risk,” effectively banning its tools from use across military systems and among contractors. The designation followed a dispute over operational guardrails governing how the military could deploy its AI capabilities.
Despite the ban, Anthropic’s technology remains highly regarded within defense circles. Pentagon staff, contractors, and former officials told Reuters they have been reluctant to phase out its tools, citing their performance advantages over competing systems. Orders are currently in place to remove Anthropic products within six months.
Faster Integration, Broader Reach
Since the fallout, the Pentagon has significantly accelerated its onboarding process for new AI vendors. Companies report that gaining approval to operate on classified systems now takes less than three months—a sharp reduction from the previous timeline of up to 18 months.
This rapid shift underscores the military’s growing dependence on artificial intelligence. AI systems are already embedded in a wide range of defense activities, from mission planning and logistics coordination to targeting and operational analysis. By expanding the pool of providers, the Pentagon says it aims to avoid “vendor lock,” a term suggesting concern over excessive dependence on a single dominant supplier.
Its flagship AI system, GenAI.mil, has seen rapid adoption, with more than 1.3 million Defense Department personnel using the platform within just five months of its launch.
Meanwhile, Google has reportedly finalized a deal allowing the Pentagon to deploy its AI models for classified work—further signaling how deeply commercial AI is becoming embedded in national security infrastructure.
Concerns Persist Over Anthropic
Pentagon Chief Technology Officer Emil Michael reaffirmed concerns about Anthropic, stating in a CNBC interview that the company “remained a supply-chain risk.” However, he distinguished those concerns from the broader implications of Anthropic’s advanced model known as Mythos.
Describing Mythos as “a separate national security moment,” Michael pointed to growing unease among policymakers and corporate leaders over the model’s potential to significantly enhance cyberattack capabilities. Its powerful features have sparked debate about how such tools could be weaponized if left unchecked.
Although several organizations have gained early access to Mythos to strengthen their cybersecurity defenses, it remains unclear whether the Pentagon is among them.
Political Signals Suggest Possible Reversal
There are signs the standoff may not be permanent. U.S. President Donald Trump recently suggested that Anthropic is “shaping up” in the eyes of his administration, hinting that the company could eventually be reinstated as a defense partner.
For now, however, the Pentagon appears focused on widening its AI ecosystem—betting that a more diverse lineup of providers will deliver both technological edge and strategic resilience in an increasingly AI-driven battlefield.