Google negotiates Pentagon deal to deploy Gemini AI in classified military environments

Google and U.S. Department of Defense negotiate breakthrough classified AI deployment

Google is negotiating with the U.S. Department of Defense to deploy Gemini AI models in classified settings, marking a significant expansion of the tech giant's military cooperation1

1

2

2

. The discussions center on enabling the Pentagon to use Google's AI for all lawful purposes while addressing ethical concerns that have derailed similar partnerships. According to sources with direct knowledge, the talks involve adding GPU racks to Google Distributed Cloud and, for the first time, deploying tensor processing units (TPUs) inside accredited classified environments1

1

.

Google Distributed Cloud received DoD Impact Level 6 authorization for Secret classified data in May 2025, alongside existing Top Secret accreditations that nominally make Gemini AI and Vertex AI available at those classification levels1

1

. However, relatively little infrastructure currently exists inside the accredited boundary to run classified workloads at scale, creating a gap that these negotiations aim to close.

Tensor processing units deployment marks technical milestone for classified AI

The proposed deployment of TPUs for classified military purposes represents a technical breakthrough, as these custom AI accelerators have never been used in secure government environments before1

1

. TPUs currently run the bulk of Gemini training and inference in Google's commercial cloud, making their deployment on the classified side essential for any Gemini rollout beyond small-scale pilots. In the short term, Google plans to add racks of GPUs to Google Distributed Cloud while working on the parallel workstream to enable TPUs.

Google pushes ethical safeguards on domestic mass surveillance and fully autonomous weapons

During negotiations, Google has proposed contract language prohibiting its AI from being used for domestic mass surveillance or fully autonomous weapons without appropriate human oversight1

1

2

2

. These terms mirror the agreement OpenAI struck with the Pentagon earlier this year, which OpenAI CEO Sam Altman asked the Pentagon to extend to all AI vendors on the same terms. The restrictions on AI without human oversight have become a central tension point in government contracts with AI companies.

These same issues broke the Pentagon's negotiations with Anthropic in February1

1

3

3

. After Anthropic declined to drop the restrictions, the Pentagon designated the company a supply chain risk and began a six-month phase-out of Claude from government systems. A federal judge has since called this designation "Orwellian" while declining to stay the ruling in one of Anthropic's two ongoing lawsuits.

Google's government partnerships expand despite Project Maven legacy

This deal marks a dramatic shift from Google's past resistance to military cooperation. In 2018, Google decided not to renew a contract with the Pentagon for Project Maven after thousands of employees voiced opposition to the company's involvement in processing images and videos for intelligence gathering3

3

. Employees feared the partnership could lead to using AI for lethal purposes.

The company has gradually reversed its stance since then. In 2022, Google Cloud established Google Public Sector to sell products to government customers, hiring former government workers and military veterans to staff the unit3

3

. According to an internal strategic plan, Google Public Sector targeted roughly $6 billion in bookings for 2025 through 2027, with $2 billion coming from defense1

1

. This remains far smaller than the over $400 billion in revenue Google generated in 2025.

Google trails rivals in cloud market and government contracts

Google holds roughly 14% of the total cloud market against 28% for AWS and 21% for Microsoft as of late 2025, per Synergy Research Group1

1

. The gap widens on the classified side, where both AWS and Microsoft run substantial workloads while Google doesn't. Data from the TBR Insight Center shows that Microsoft, Oracle, and AWS have been awarded virtually all hyperscaler government contracts, with AWS seeing a rise in defense-related contracts since 20243

3

.

Google lost last year's Army Next Generation Command and Control bid, but in July won a DoD AI pilot contract worth up to $200 million alongside Anthropic, OpenAI, and xAI1

1

. Gemini AI was the first model added to the Pentagon's unclassified GenAI.mil platform in December, and Google announced a separate deal last month for AI agent tooling on unclassified networks. The current negotiations build upon a previous agreement struck last November for the US military to use Gemini AI models for all lawful purposes in unclassified settings only3

3

.

Whether Google's proposed ethical restrictions will appear in the final agreement remains unclear, but the outcome will likely influence how other AI companies approach military cooperation and shape the future of classified AI deployment across the defense sector.