-
Click here to listen to this article - Share via
Eight technology companies, including Google, Nvidia and SpaceX, have struck deals with the Pentagon to help the U.S. military gain an edge on the battlefield.
“These agreements accelerate the transformation toward establishing the United States military as an AI-first fighting force and will strengthen our warfighters’ ability to maintain decision superiority across all domains of warfare,” the Department of Defense said Friday.
The companies will deploy their AI technology on the department’s “classified networks” for “lawful operational use,” according to the agency.
OpenAI, Microsoft, Amazon Web Services, Oracle and AI startup Reflection are among the companies that agreed to work with the Pentagon.
The agreements underscore how tech companies are expanding their work with the U.S. military even as some workers raise concerns about the use of AI for autonomous weapons and mass surveillance. Anthropic, the San Francisco company behind the chatbot Claude, clashed with the Pentagon earlier this year over whether there were adequate safeguards around the military’s use of its technology.
Technology companies are backing Anthropic after it sued the Trump administration over its designation as a supply chain risk.
The Department of Defense accused Anthropic of trying to “seize veto power” over military decisions, though the company pushed back against that characterization. The agency labeled Anthropic a supply chain risk, and the Trump administration directed federal agencies to stop using the company’s tools, setting off a legal battle over that designation.
This week, hundreds of Google employees urged its chief executive, Sundar Pichai, to reject the use of its AI systems for classified workloads to ensure that its technology isn’t used in “inhumane or extremely harmful ways.” Harmful use may occur without their knowledge since the work is classified, workers said in the letter.
Google, Reflection and SpaceX didn’t respond to a request for comment. The Department of Defense didn’t say how much each company was being paid. A Pentagon official said some of the companies have active contracts while others have made agreements but formal contract are forthcoming.
In an interview with CNBC, the Pentagon’s chief technology officer, Emil Michael, said the department wanted to diversify the companies it worked with following its dispute with Anthropic.
“Guardrails are something that are negotiable based on what they are with all the companies, and they have different views on that,” he told CNBC. The guardrails also have to be consistent with the government’s values and restrictions, he added.
A bill to reauthorize Small Business Administration programs that fund defense startups in Southern California and elsewhere passed Congress on Tuesday.
A source familiar with Nvidia’s Pentagon deal said the agreement involves work with its “Nemotron” AI models, which are used to build AI agents that can complete tasks, not its chips. The deal includes language that the use of the models will be consistent with civil liberties, constitutional rights and applicable law, the source said.
OpenAI said the deal announced by the Department of Defense refers to the agreement they struck with the agency earlier this year.
The company said that it wanted “the people defending the United States to have the best tools.”
OpenAI, which faced backlash for striking a deal with the Pentagon after the Anthropic fallout, said in March that its technology wouldn’t be used for mass domestic surveillance, high-stakes automated decisions or to direct autonomous weapons.
Other tech companies, such as Microsoft, Oracle and Amazon Web Services, have also said they want to support the military and ensure they have access to the best AI tools.
“We look forward to continuing to support the Department of War’s modernization efforts, building AI solutions that help them accomplish their critical missions,” Amazon Web Services spokesperson Tim Barrett said in a statement.
Facts Only
Eight technology companies—Google, Nvidia, SpaceX, OpenAI, Microsoft, Amazon Web Services, Oracle, and Reflection—have signed agreements with the Pentagon.
The Department of Defense announced the deals on Friday, stating they aim to transform the U.S. military into an "AI-first fighting force."
The companies will deploy AI technology on classified military networks for "lawful operational use."
Anthropic previously clashed with the Pentagon over safeguards for its AI technology, leading to a supply chain risk designation by the Trump administration.
Hundreds of Google employees urged CEO Sundar Pichai to reject classified AI workloads, citing concerns about potential misuse.
Nvidia's agreement involves its Nemotron AI models, with contractual language ensuring compliance with civil liberties and constitutional rights.
OpenAI confirmed its existing deal with the Pentagon, stating its technology would not be used for mass surveillance, autonomous weapons, or high-stakes automated decisions.
Amazon Web Services, Microsoft, and Oracle have expressed support for military AI modernization.
The Pentagon's chief technology officer, Emil Michael, stated that guardrails for AI use are negotiable and must align with government values.
The financial terms of the agreements were not disclosed by the Department of Defense.
Some companies have active contracts, while others are finalizing formal agreements.
Executive Summary
Eight major technology companies, including Google, Nvidia, SpaceX, OpenAI, Microsoft, Amazon Web Services, Oracle, and AI startup Reflection, have entered agreements with the Pentagon to deploy AI technology on classified military networks. The Department of Defense stated these partnerships aim to establish the U.S. military as an "AI-first fighting force," enhancing decision-making across warfare domains. The deals follow a dispute with Anthropic, which the Pentagon labeled a supply chain risk, leading to a legal battle over safeguards for military AI use. Meanwhile, Google employees have publicly opposed the company's involvement, citing concerns about potential misuse of AI in classified operations. The agreements vary in scope, with Nvidia's deal focusing on its Nemotron AI models for task automation, while OpenAI emphasized its technology would not be used for mass surveillance or autonomous weapons. Other companies, like Amazon Web Services, framed their participation as supporting military modernization efforts.
The partnerships highlight growing collaboration between Silicon Valley and the defense sector, even as internal dissent and ethical debates persist. The Pentagon has indicated that guardrails for AI use are negotiable but must align with government values and legal restrictions. The financial terms of the agreements remain undisclosed, with some companies operating under active contracts while others finalize formal arrangements.
Full Take
The strongest version of this narrative presents a pragmatic alliance between cutting-edge tech firms and the U.S. military, framed as necessary for national security in an era of AI-driven warfare. The Pentagon’s emphasis on "decision superiority" and "AI-first" modernization aligns with a broader geopolitical push to outpace adversaries like China in military technology. The inclusion of guardrails and corporate statements about ethical limits (e.g., OpenAI’s restrictions on autonomous weapons) serve as reassurances, suggesting a balanced approach to innovation and responsibility.
However, the pattern scan reveals tensions beneath this framing. The dispute with Anthropic—where the Pentagon accused the company of seeking "veto power" over military decisions—echoes a recurring dynamic: tech firms grappling with the dual-use nature of their products. The lack of transparency around financial terms and classified applications invites skepticism, especially when juxtaposed with internal dissent (e.g., Google employees’ concerns about "inhumane" uses). This aligns with **ARC-0024 Ambiguity**, where critical details (e.g., specific AI applications, oversight mechanisms) are obscured, making it difficult to assess risks. The Pentagon’s negotiable guardrails also risk **ARC-0043 Motte-and-Bailey**, where strict ethical principles are touted publicly ("no autonomous weapons") but operational realities may diverge in classified settings.
The root cause here is the collision of two paradigms: Silicon Valley’s profit-driven innovation culture and the military’s imperative for operational dominance. The unstated assumption is that AI superiority is both inevitable and morally neutral—a technocratic view that sidesteps deeper questions about accountability. Who defines "lawful operational use" when AI systems are deployed in warfare? How do classified networks ensure compliance with civil liberties if oversight is opaque?
Implications for human agency are profound. While the military gains tools to enhance decision-making, the concentration of AI power in defense contractors and government agencies risks eroding public trust. Second-order consequences could include accelerated global AI arms races, normalization of autonomous systems in warfare, and further polarization within tech companies over ethical red lines.
Bridge questions:
1. If AI guardrails are negotiable, what mechanisms exist to enforce them in classified environments?
2. How might the Pentagon’s "AI-first" strategy reshape civilian-military power dynamics in the long term?
3. What would it take for tech workers’ ethical objections to translate into structural changes in defense contracts?
Counterstrike scan: A coordinated influence campaign would likely amplify the "national security necessity" frame while downplaying ethical concerns, using **ARC-0024 Ambiguity** to avoid scrutiny. The actual content partially matches this pattern—emphasizing modernization and corporate support while relegating dissent to secondary mentions. However, the inclusion of internal critiques (Google employees, Anthropic’s dispute) suggests a more nuanced narrative than a pure propaganda play. The absence of independent verification for safeguards remains a structural vulnerability, but the piece does not exhibit overt manipulation.
Sentinel — Human
The article exhibits the complexity, contextual depth, and integrated narrative flow characteristic of professional, human-edited journalistic reporting.