Building America’s AI Arsenal

Building America’s AI Arsenal: The Case for a National Computing Reserve

A Category 5 hurricane barrels toward the Gulf Coast. Unlike Helene and Milton, this time AI leads our response.

Within hours, it precisely predicts the storm’s path, identifies vulnerable areas, and optimizes evacuation routes. Emergency responders are deployed with unprecedented efficiency. Lives are saved. Damage is minimized. This isn’t just better disaster management—it’s a revolution in saving lives.

On today’s battlefield, commanders need AI-powered tools to process vast amounts of data and make split-second decisions. From processing sensor data to coordinating multi-domain operations, military leaders require unprecedented computing power at the tactical edge.

But to make this vision a reality, we need to act now. Three years ago, the White House Office of Science and Technology Policy proposed a National Strategic Computing Reserve—a coalition ready to contribute massive computational power in times of national need. It was a good idea then. Today, it’s an urgent necessity.

Why? Because in those three years, our need for compute power—the raw processing capability that drives our digital world—has exploded. AI has evolved at a blistering pace, turning once-futuristic capabilities like real-time language translation and complex strategic decision-making into everyday realities.

The numbers are staggering. Since 2018, the computational requirements for training large language models have doubled every six months. By 2023, the largest AI training runs were using over a million times more compute than a decade ago. And that’s just for training. Running these models across countless real-time applications pushes demand even higher, straining our computational infrastructure to its limits.

This isn’t just about keeping Silicon Valley on the cutting edge. It’s about our national capacity to respond to crises. Imagine AI rapidly identifying treatments during a pandemic, neutralizing cyber threats in real-time, or simulating economic policies to stabilize markets during a shock. These capabilities could save lives and protect our nation—but only if we have the necessary compute power ready to deploy across the full spectrum of operations, from enterprise data centers to tactical edge environments.

Yet the United States lacks a coordinated strategy to ensure sufficient compute power when it matters most. As AI becomes integral to our national infrastructure and defense systems, having this capacity isn’t just an economic advantage—it’s a strategic imperative.

A National Strategic Computing Reserve would solve this. Think of it as a digital-age parallel to the Strategic Petroleum Reserve. In peacetime, it would accelerate scientific research and drive innovation. During crises, it could be rapidly mobilized to address urgent national needs.

The Department of Defense understands this imperative. At the tactical level, military units are already developing AI-enabled decision support tools that can function in denied or disrupted environments. These initiatives demonstrate how a strategic compute reserve could work: standardized platforms, interoperable systems, and computing power available when and where it’s needed most.

What would this reserve look like? Imagine a network of high-performance computing centers across the country, with both cloud-based and edge computing capabilities, ready to support multi-domain operations and be redirected to national priorities in times of crisis.

Partnerships with leading tech companies could provide cloud computing resources on demand. We’d need to invest in next-generation AI chips and quantum computing technologies to stay ahead of the curve. And crucially, we’d need training programs to ensure a skilled workforce capable of leveraging these resources effectively.

Critics might argue that the private sector can meet our computing needs. But national emergencies demand a dedicated, rapidly deployable infrastructure. Relying solely on private companies—with their potentially conflicting global interests—could leave us vulnerable in a crisis.

Concerns about energy consumption are valid, but the strategic value of this computing power far outweighs these issues. In fact, this initiative could drive innovations in energy-efficient computing and accelerate our shift to renewables.

The cost of creating a National Strategic Computing Reserve might seem high, but the price of inaction is far steeper. This isn’t just about keeping pace; it’s about maintaining our lead in a high-stakes global race. Being computationally outgunned could have dire consequences for our national security and economic future.

The global AI race has reached a critical juncture. China now leads in AI research output, with five of its institutions topping worldwide rankings. While the U.S. maintains an edge in research quality, China’s AI patent filings quadrupled ours in 2022. Even Russia is ramping up its capabilities. The stakes are clear: whoever leads in AI will shape its future development.

Three years have passed since this idea was first proposed, and our advantage in compute resources has narrowed daily. We can’t afford to lose more time. The invisible arms race of the 21st century is well underway, and the victors will be those who can harness AI at unprecedented scale and speed.

America has always thrived when it thinks big and acts boldly, and now is the time for such action. A National Strategic Computing Reserve is exactly the kind of forward-thinking initiative we need to secure our place in the AI-driven future. But we must move swiftly; the next crisis—or the next AI breakthrough—won’t wait for us to catch up.

The choice before us is clear: lead in the age of AI, or risk being left behind. The military’s current digital transformation efforts show us the way forward. By embedding technical expertise at every level and ensuring access to sufficient compute power, we can enable rapid, informed decision-making in even the most challenging scenarios. A National Strategic Computing Reserve would extend these capabilities across our entire national security infrastructure.


Ben Van Roo is the co-founder and CEO of Yurts, a generative AI company partnering with the US Department of Defense to advance mission-critical systems. He holds a PhD in Operations Research and has significant experience developing AI solutions for defense and national security applications.