Eight years ago, thousands of Google employees signed a petition demanding the company abandon Project Maven, a Pentagon initiative that used artificial intelligence to analyze drone surveillance footage. Google backed down. It let the contract lapse. It published a set of AI principles that seemed to draw a bright line between its technology and lethal military applications.
That line has been steadily erased.
According to a detailed report from The New York Times, Google has aggressively expanded its defense and intelligence work over the past several years, building a portfolio of Pentagon contracts that dwarfs anything it pursued during the Maven era. The company now provides cloud computing infrastructure, AI-powered data analysis tools, and cybersecurity services to multiple branches of the U.S. military and intelligence community. And it’s done so with far less internal resistance than the firestorm that engulfed the company in 2018.
The shift tells a broader story about Silicon Valley’s evolving relationship with Washington — one driven by geopolitical anxiety, enormous contract dollars, and a tech workforce that has, by attrition or acquiescence, largely stopped pushing back.
From Maven’s Ashes to Multi-Billion-Dollar Ambitions
Google’s retreat from Project Maven was, at the time, seen as a watershed moment. Here was one of the world’s most powerful technology companies bowing to the moral objections of its own engineers. The company’s AI principles, published in June 2018, stated that Google would not design or deploy AI for weapons or technologies whose purpose was to cause injury. It felt definitive.
It wasn’t.
The principles contained enough ambiguity to allow substantial military work. Google could still build AI for cybersecurity. For logistics. For intelligence analysis that didn’t directly guide weapons. And critically, cloud infrastructure — the backbone of modern military computing — was never off the table. Within two years of the Maven controversy, Google was bidding on the Pentagon’s Joint Enterprise Defense Infrastructure contract, a cloud computing deal worth up to $10 billion. It lost that bid to Microsoft, but the pursuit itself signaled where the company was heading.
By 2024, Google Cloud had secured contracts with the Department of Defense, the intelligence community, and agencies across the national security apparatus, as The New York Times documented. The company established a dedicated public sector division, hired executives with deep Pentagon ties, and opened offices near military installations. Thomas Kurian, the CEO of Google Cloud, made defense work a strategic priority, viewing it as essential to competing with Amazon Web Services and Microsoft Azure, both of which had built formidable government businesses.
The numbers are significant. While Google doesn’t break out defense revenue separately, analysts estimate its government and defense cloud contracts now generate several billion dollars annually. That’s a fraction of Google’s overall revenue — parent company Alphabet reported $350 billion in 2025 — but it represents one of the fastest-growing segments of Google Cloud’s business.
So what changed? Multiple things, simultaneously.
The geopolitical environment shifted dramatically. Russia’s invasion of Ukraine in 2022 created a sense of urgency around military modernization. China’s rapid advances in AI-powered military systems alarmed Pentagon officials and congressional leaders alike. The argument that American tech companies had a patriotic obligation to support national defense gained traction — not just in Washington, but inside Silicon Valley itself.
The workforce changed too. Many of the most vocal opponents of military work left Google during and after the Maven controversy. Some departed for smaller companies or academia. Others were part of broader layoffs. The employees who remained, or who joined afterward, were less likely to view defense contracts as a red line. A tighter job market in tech, following the mass layoffs of 2023 and 2024, further dampened internal dissent. Hard to organize a walkout when you’re worried about your job.
Google’s leadership also became more assertive. Sundar Pichai, who had appeared uncomfortable during the Maven backlash, grew more explicit about the company’s willingness to work with government and military clients. In internal communications reviewed by The New York Times, executives emphasized that Google’s AI principles permitted defense work that was defensive in nature, that protected service members, or that supported cybersecurity operations. The framing was deliberate: this wasn’t about building killer robots. It was about protecting the country.
Critics aren’t buying it. “The distinction between offensive and defensive AI in a military context is essentially meaningless,” said Lucy Suchman, a professor emerita at Lancaster University who has studied the ethics of AI in warfare, in comments to The Times. Cloud infrastructure that powers intelligence analysis can just as easily support targeting decisions. An AI model trained to identify objects in satellite imagery doesn’t know — or care — whether its output is used for humanitarian mapping or strike planning.
The Competitive Pressure That Rewired Silicon Valley’s Conscience
Google isn’t acting alone. The entire major tech industry has moved toward defense work with striking speed.
Microsoft has been the most aggressive, winning the $10 billion JEDI contract (later restructured as the Joint Warfighter Cloud Capability) and supplying augmented reality headsets to the Army through its IVAS program. Amazon Web Services runs the CIA’s classified cloud infrastructure and has expanded across the intelligence community. Even companies that once kept their distance — Salesforce, Oracle — have built dedicated government divisions.
But the most telling development may be the rise of defense-focused AI startups that have no Maven-era baggage at all. Companies like Anduril Industries, Palantir Technologies, and Shield AI have built their entire identities around serving the military. Palantir, founded by Peter Thiel, has become one of the Pentagon’s most important AI contractors, with its stock price surging as defense spending on AI accelerated. Anduril, led by Palmer Luckey, designs autonomous drones and surveillance systems with an explicitly pro-military ethos.
These companies created competitive pressure that Google couldn’t ignore. If Google refused to provide AI tools to the Pentagon, Palantir or Anduril would — and they’d build those tools on Amazon’s or Microsoft’s cloud infrastructure. Google wouldn’t just lose the direct defense revenue. It would lose the underlying cloud business that came with it. For a company trying to close the gap with AWS and Azure, that was unacceptable.
There’s also the matter of political reality. The Biden administration pushed hard to integrate commercial AI into defense applications, launching initiatives like the Pentagon’s Chief Digital and Artificial Intelligence Office (CDAO) and the Replicator program, which aimed to field thousands of autonomous systems. The Trump administration, now in its second term, has accelerated those efforts while making clear that tech companies unwilling to work with the military risk regulatory and political consequences.
Congressional leaders from both parties have been blunt. Senator Mark Warner, the Virginia Democrat who chairs the Senate Intelligence Committee, has repeatedly argued that American AI superiority depends on collaboration between the private sector and the military. Republican hawks have gone further, suggesting that companies refusing defense work should lose access to government research funding or face antitrust scrutiny.
Google got the message.
The company’s 2018 AI principles haven’t been formally rescinded, but they’ve been reinterpreted so broadly that they impose few practical constraints. A Google spokesperson told The New York Times that the company continues to evaluate all contracts against its principles and that it has declined certain projects. But the spokesperson declined to specify which projects were rejected or how the review process works. The principles now function less as a hard boundary and more as a public relations framework — a way to signal ethical seriousness without meaningfully limiting business opportunities.
Inside Google, the defense push has been accompanied by a cultural shift that extends beyond military contracts. The company has become less tolerant of internal activism generally. After firing several employees involved in protests over a cloud contract with the Israeli government in 2024, Google sent a clear signal: dissent on business decisions would carry professional consequences. The open, freewheeling culture that once defined the company — where engineers could question leadership on internal forums and organize petitions without fear — has given way to something more corporate. More controlled.
Not everyone has accepted this quietly. Some current and former employees have spoken to journalists and advocacy organizations about their concerns. The Tech Workers Coalition, an activist group, has criticized Google’s defense expansion and called for greater transparency about military AI applications. But these voices are smaller and less influential than the coalition that forced the Maven retreat. The balance of power inside the company has shifted decisively toward leadership.
The financial incentives are enormous. Federal spending on AI and cloud computing for defense is projected to exceed $30 billion annually by 2028, according to estimates from research firm Govini. That’s money Google, Amazon, and Microsoft are all chasing aggressively. And unlike consumer markets, where growth has slowed, government contracts offer long-term revenue streams with high switching costs. Once a military branch builds its infrastructure on Google Cloud, migrating to a competitor is extraordinarily expensive and disruptive.
Google has also been investing in specialized capabilities designed specifically for defense and intelligence clients. Its Distributed Cloud product, which allows government agencies to run Google Cloud services in classified environments and at the tactical edge — on military bases, aboard ships, even in forward-deployed locations — has become a key selling point. The company has obtained high-level security certifications, including FedRAMP High and Impact Level 5 authorization, that allow it to handle classified workloads.
These aren’t trivial technical achievements. They represent years of investment and engineering effort, undertaken with the explicit goal of making Google indispensable to the national security establishment.
What Comes Next — And What’s at Stake
The implications extend well beyond Google’s balance sheet.
As AI becomes more deeply embedded in military decision-making, questions about accountability, transparency, and the laws of armed conflict become urgent. Who is responsible when an AI system misidentifies a target? What oversight exists when algorithms process intelligence data that informs lethal operations? How do international humanitarian laws apply to autonomous systems that operate at machine speed?
These aren’t hypothetical concerns. The Pentagon is actively deploying AI in operational contexts — from intelligence analysis in the Middle East to autonomous drone swarms being tested for Pacific contingencies. Google’s technology may not directly control weapons today, but the infrastructure it provides underpins systems that do. The company’s insistence that it only works on “defensive” applications becomes harder to sustain as the line between defense and offense blurs in modern warfare.
There’s a historical parallel worth considering. In the mid-20th century, companies like IBM, AT&T, and General Electric built deep partnerships with the U.S. military that lasted decades and shaped both the companies and the country. Those relationships produced technological breakthroughs — the internet itself grew out of a Pentagon research project — but also entangled corporations in morally complex situations, from nuclear weapons development to surveillance programs.
Google and its peers are walking a similar path. The technology is different. The stakes may be higher. AI systems can make decisions and take actions at speeds no human can match, creating risks of escalation and error that didn’t exist in previous eras. And unlike the Cold War defense contractors, today’s tech giants possess intimate knowledge of billions of people’s lives — their communications, movements, relationships, and habits. The convergence of that commercial data capability with military applications raises questions that existing legal and ethical frameworks weren’t designed to answer.
For Google specifically, the defense expansion represents a definitive break from the idealism that once characterized the company. “Don’t be evil,” the informal motto that defined early Google, was quietly replaced by Alphabet’s more anodyne “Do the right thing” in 2015. But even that softer formulation sits uneasily alongside contracts that support military operations in conflict zones.
Google’s leaders would argue — and have argued — that the right thing is precisely to ensure American military superiority over authoritarian rivals. That if Google doesn’t provide AI to the Pentagon, China’s tech champions will provide it to the People’s Liberation Army, and the consequences of that asymmetry could be catastrophic. It’s a compelling argument. It’s also one that can justify almost anything.
The employees who protested Maven in 2018 understood something that remains true: once a technology company becomes deeply integrated into the military apparatus, extracting itself becomes nearly impossible. Contracts beget contracts. Dependencies deepen. The revenue becomes too significant to walk away from. And the political cost of withdrawal — in an era of great-power competition — becomes prohibitive.
Google has made its choice. The company that once recoiled from a relatively modest drone imagery contract now builds the cloud infrastructure that the Pentagon depends on for AI-driven warfare. The protests are over. The principles remain on the website. And the contracts keep getting bigger.
WebProNews is an iEntry Publication