The company that does not manufacture missiles or tanks, but makes the brains that control them

In Santa Clara County, California, a green flag still flies. In appearance, it is just the emblem of another Silicon Valley technology company. However, behind that insignia hides a corporation that today moves the foundations of global power. Their products don’t shoot, but they allow others to do so with a precision impossible just a decade ago. Its chips process intelligence, train drones, simulate combat scenarios and coordinate decisions in the most sophisticated command centers on the planet.

For years, his name has been associated with the gamers and video games. Currently its presence reaches from Pentagon laboratories to NASA space missions. The United States Government has included it on the list of essential suppliers of the Defense Production Act, a status reserved for companies whose activity is considered strategic for national security. Since then, its technology no longer just accelerates graphics, it also powers the digital infrastructure of the US defense.

The road to that position has not been abrupt. It all started with research contracts with the Defense Advanced Research Projects Agency (DARPA), which was seeking faster processors for autonomous systems and flight simulators. Agreements were later reached with the Department of Energy and NASA. In May, the MITER Corporation organization, a frequent collaborator of federal agencies, announced the construction of a supercomputer with 256 units of this company to train artificial intelligence (AI) models in national security environments. The project, valued at about 20 million dollars, confirmed what many already anticipated, that the power of the 21st century will be measured in computing capacity.

The identity of the company is no longer a mystery when you look at its scale. It is Nvidia, popularly known for being the most valuable company on the planet with a capitalization close to $5 trillion. It is the “pretty girl” of the world stock market that has turned chips into a tool of strategic power. What once served to improve a video game today guides satellites, drones and defense simulators.

Nvidia’s new battlefield

But its influence is not limited to software. This year, together with BlackRock, Microsoft and the Abu Dhabi fund MGX, it closed the purchase of Aligned Data Centers for $40 billion. The operation opens the door to control of the physical spaces where the most powerful AI algorithms in the world are processed. Each of those centers houses thousands of Nvidia drives intended for scientific simulations, strategic surveillance and defense model training. At the same time, the company has started domestic production of its next-generation chips.

Blackwell cards will be manufactured in plants in Arizona and Texas with a investment of more than 50 billion dollars. The initiative is part of the technological sovereignty strategy promoted by Washington and seeks to reduce dependence on Asia in full rivalry with China. For the US Government, guaranteeing the supply of these processors is as important as ensuring access to energy. Its footprint also extends to Europe, where the MareNostrum 5 supercomputers in Barcelona and Leonardo in Bologna use its architecture within the EuroHPC program.

These systems, designed for civil research, are prepared for defense and security applications. On the other hand, India has signed an agreement with the company to deploy AI innovation centerswhile Saudi Arabia and the United Arab Emirates finance, together with it, data complexes that will serve as infrastructure for the development of large-scale models. The scope of its technology has reconfigured the rules of the game.

According to Bloomberg Intelligencealmost a third of global defense spending linked to AI depends directly or indirectly on its hardware. In the United States, its processors power the systems of the JADC2 program, the network that connects all branches of the military in a unified digital structure. In this environment, each drone or military simulator uses algorithms optimized for its GPUs, capable of processing decisions in milliseconds.

Military contracts

Export restrictions imposed by Washington on the most advanced chips have further cemented its role as a security ally. The H100 and H200 models, essential for training large-scale neural networks, can only be sold outside the country with a specific license. The Department of Commerce justifies this measure for defense reasons and considers it a national protection tool.

An analysis by the Brookings Institution reveals that AI contracts related to the US military increased from less than $200 million in 2022 to more than $550 million in 2023, helping to understand the environment in which Nvidia operates. The numbers complete the picture. In 2024 the company had a turnover of 186 billion dollars, 125% more than the previous year, and obtained a net profit of 80 billion.

The portion of income from public and military contracts exceeds 14%, according to Financial Times. The rest comes from the civilian market, where its chips dominate the cloud of Amazon, Google and Microsoft and process everything from financial algorithms to autonomous driving systems. The result is an unprecedented balance between private capital and public function. Nvidia continues to present itself as a civil company, but its technology has been integrated into the Western security infrastructure. It does not manufacture missiles or tanks, but it does manufacture the brains that control them, and in an economy that runs on algorithms, that role places it at the very center of military power.