ON Jan 17, 1961, Dwight D. Eisenhower sat before a television camera to deliver his farewell address as the president of the United States. Instead of a simple ceremonial goodbye, the former five-star general, who had commanded Allied forces in Europe during World War II and later presided over America’s Cold War security state, issued a warning.
America, he said, had built something new: a permanent arms industry closely tied to military power, scientific research, political influence and state spending. His fear was not the production of weapons alone, but the creation of a permanent economy in which war became a source of contracts, influence, research budgets and political power.
Eisenhower called it the “military-industrial complex” and warned that its influence could endanger liberty and democratic life.
More than six decades later, that warning is beginning to return in a form Eisenhower could not have imagined. The old factories produced aircraft, missiles, tanks and submarines. The new ones look very different: cloud systems, AI models, chips, data centres, satellite networks and battlefield software. Together, they form the computational backbone on which states are beginning to depend for intelligence, prediction and military decision-making.
Democratic states may begin to depend on privately owned systems they cannot fully audit, regulate or understand.
This is why the Pentagon’s recent declaration that the US military must become an “AI-first” fighting force should not be read as a routine technology announcement. It is a political and historical marker.
The US Department of Defence has entered into agreements with leading frontier AI and technology companies, including OpenAI, Google, Nvidia, Microsoft, Amazon Web Services, Oracle and SpaceX, to deploy advanced AI capabilities on classified military networks. Its stated purpose is to accelerate the transformation of the US military into an AI-first force and strengthen “decision superiority” across domains of warfare.
In plain terms, Silicon Valley has moved from the edges of the national security state into its core machinery.
This is the new military-industrial complex. But it is more difficult to see because it does not look like the old one. A weapons factory announces itself as part of the war economy. A cloud company does not. A missile manufacturer is easily understood as a defence actor. An AI lab, a chipmaker, a satellite internet company, or a data analytics firm can present itself as civilian, commercial, educational, or even humanitarian, depending on the customer and the contract.
That is precisely what makes this moment so important. The same companies that host the world’s data and communications, mediate information flows and build consumer AI assistants are now becoming central to war and national security itself. Their power now extends into the systems states use for surveillance, intelligence, targeting and military command.
Thus, the same AI tool that helps a student write, a journalist summarise, or a company analyse data can also help a military process intelligence. In the age of AI, civilian infrastructure and military power increasingly run on the same systems.
For democracies, this raises a profound question. What happens when the corporations that mediate reality for billions of people also become indispensable to the machinery of war? The problem is not only that companies may profit from conflict. That is an old problem. The deeper problem is that democratic states may begin to depend on privately owned systems they cannot fully audit, regulate or understand.
Imagine one company hosting vast amounts of the world’s data while also building surveillance and military technologies. What does that mean for countries outside the US, whose citizens, businesses, and institutions may already depend on its systems?
This connects directly to the argument I made earlier in these pages (‘The age of AI empires’). The real ‘AI race’ is about chips and power. This also explains why energy has returned to the centre of geopolitics. Data centres need enormous, continuous power, and the International Energy Agency now estimates that their global electricity consumption could roughly double from 485 terawatt-hours in 2025 to 950 terawatt-hours by 2030.
For states still struggling to build their own technological capacity, the warning is clear. Dependence is no longer limited to imported machines or software. It now extends to the platforms where citizens speak, the cloud systems where businesses operate, the AI tools newsrooms use, and the systems through which young people increasingly learn about the world. Much of this infrastructure is built, priced, governed and politically shaped elsewhere.
This dependency has consequences for free speech and fundamental rights. If platforms shape visibility, then speech is not only a constitutional matter but also an infrastructural matter. If AI systems shape access to knowledge, then education, journalism, and public debate are affected by systems whose rules may not be transparent. If the same companies are also defence partners of powerful states, smaller countries must ask whether their information ecosystems can remain neutral, open and locally accountable.
For countries outside the US, detaching from global technology is neither realistic nor desirable. The real issue is whether we understand the dependence we are entering into. If speech, knowledge, business, journalism and governance increasingly pass through systems built elsewhere, then digital capacity is no longer a side issue. It becomes part of sovereignty itself. Any state that cannot build, audit or properly negotiate with digital infrastructure will eventually find that many of its choices have already been made for it.
Eisenhower’s warning was against misplaced power. In his time, that power emerged from the fusion of arms, industry and the state. In ours, it is emerging from the fusion of AI, cloud, data, energy and war. That is why the question before democracies is no longer whether Silicon Valley will shape the future of public life. It already does. The question now is whether it will also shape the future of war before democratic societies, especially those outside the centres of technological power, have found the language, laws and leverage to respond.
The writer is the founder of Media Matters for Democracy.
Published in Dawn, May 18th, 2026
