Google and the Pentagon sign a $200 million AI military contract to implement Gemini, 600 employees protest unsuccessfully

Google has officially signed a $200 million AI military contract with the U.S. Department of Defense, allowing the Pentagon to access Google’s commercial models via API for “legitimate government purposes.” Over 600 Google employees, including DeepMind researchers, signed an open letter demanding CEO Pichai reject this deal, but ultimately, the outcome did not change.
(Background: 600 Google employees jointly protest: Reject Gemini’s entry into Pentagon’s classified military network)
(Additional context: Trump hints that the Department of Defense is very likely to reach an AI cooperation agreement with Anthropic, praising: they are making progress)

Table of Contents

Toggle

• Contract details: API access, no custom development
• The gap left by Anthropic’s departure
• 600 signatures: Can protests change anything?

Earlier this week, Google’s AI military contract with the U.S. Department of Defense officially took effect. On the same day, over 600 Google employees’ signed protest letter was also delivered to CEO Sundar Pichai’s inbox.

This is not a coincidence but a snapshot of a long-standing internal tug-of-war within Google: the company’s direction, employees have differing opinions, but ultimately, the decision-making power always lies with management.

Contract details: API access, no custom development

Bloomberg reporters confirmed the details: the contract is worth approximately $200 million, allowing the U.S. military to use Google AI systems for “any legitimate government purpose.”

A Google spokesperson emphasized in a statement that this is an API access contract: the Pentagon can connect directly to Google’s commercial models, but it does not include custom work or model development.

“We believe that providing API access to our commercial models under industry-standard terms, including running on Google infrastructure, is a responsible way to support national security.”

Simply put, API access means the Pentagon’s systems can directly “call” Google’s AI models to obtain analysis results, but Google does not build dedicated tools from scratch for the military. Google also announced:

“The company remains committed to a consensus between the public and private sectors: AI should not be used for large-scale domestic surveillance or autonomous weapons without proper human oversight.”

The gap left by Anthropic’s departure

The timing of this contract is hard to understand without considering a larger background.

Earlier this year, the Pentagon’s collaboration with Anthropic broke down. They are now trying to expel Anthropic and its Claude AI tools from the U.S. defense supply chain and actively seek new tech giants as partners. Google’s signing of this contract is widely seen as filling the void left by Anthropic’s exit.

Read more: Pentagon confronts Anthropic! Fully opens Claude for military use “or else terminate contract”

For Google, this is also a reaffirmation of its strategic direction. In 2018, Google employees protested the company’s involvement in the Pentagon’s Project Maven (a military project using AI to analyze drone footage), described by Bloomberg as “the core of modern American warfare.” That protest led Google to not renew the contract and publicly commit to not developing weapons or other potentially harmful technologies.

However, by 2025, Google quietly removed some restrictive language from its AI ethics policies. This new military contract is the first major outcome after policy liberalization.

Interestingly, Google also quietly withdrew from a $100 million Pentagon project during the same period. The project aimed to develop voice-controlled autonomous drone swarms. Internal files obtained by Bloomberg show the withdrawal was due to internal ethical reviews; but Google’s official statement was “lack of resources,” reflecting the company’s consistent ambiguous stance on military AI issues.

600 signatures: Can protests change anything?

On the day of the contract signing, over 600 Google employees, including DeepMind researchers, submitted an open letter to Pichai. The letter stated:

“We are Google employees deeply concerned about the negotiations between Google and the U.S. Department of Defense. As AI practitioners, we know these systems can concentrate power, and they do indeed make mistakes.”

The employees’ concerns focused on two points: the risk of errors in AI systems in military contexts, and the irreversible consequences of concentrated power. The protest letter clearly demanded: reject the contract. But management still decided to proceed with signing.

This is the second large-scale protest by Google employees against military AI contracts; the outcome is very different from 2018. That protest led Google to withdraw from Project Maven; this time, the protest letter and the contract almost coincided in time, but the 600 signatures did not change any business decision.

Read more: 600 Google employees jointly protest: Reject Gemini’s entry into Pentagon’s classified military network