AI is no longer an emerging technology. It’s here, and it’s becoming the bedrock of modern civilization. Just as electricity transformed the 20th century and the Internet transformed the 21st, AI is reshaping how we work, govern, and live. Soon, every major institution, from hospitals to the military, will integrate AI into their core operations, raising the stakes for the infrastructure that underpins it.
Despite this demand, our infrastructure isn’t keeping pace. In 2024, U.S. data centers used ~200 terawatt-hours of electricity, enough to power Thailand for a year. The same estimate holds that by 2028, AI power usage is predicted to reach between 165 and 326 terawatt-hours annually, enough to power 22% of U.S. households. AI workloads are pushing energy and compute systems well beyond their limits, creating an exponential demand that leaves our power grid lagging behind as it struggles to scale even incrementally.
This mismatch is more than a technical issue. As demand for AI ramps up, these bottlenecks in national energy supply and compute access will slow development across every sector, limiting its transformative potential.
The United States is leading, for now. But we are in a sprint, and China is gaining ground. Their DeepSeek model R1 rivals top-tier U.S. models. DeepSeek’s success proves that speed, scale and efficiency can radically shift the balance of global AI power. China’s AI push is well-funded, coordinated and strategic. If DeepSeek is any indication of China’s momentum, we are far behind them.
It won’t matter who leads in algorithms if the U.S. keeps treating infrastructure as an afterthought, because we’re on track to lose the platform war. The future of AI must be built on freedom, transparency, and trust, not surveillance and control. That is America’s edge—and to that we must prioritize the energy crisis it’s creating.
In this context, massive, centralized data centers are obsolete. They’re rigid, expensive, and confined to one geographic location. Even worse, they create single points of failure. If one power grid goes down or is overheated, an entire segment of the country is plunged into a technological dark age.
By contrast, decentralized systems free our potential, allowing American innovation to scale with agility. Smaller compute clusters can run near sources of localized renewable energy, such as solar, wind, or geothermal energy, or take advantage of underutilized compute power sitting idle in homes, campuses, and communities. Decentralized systems also better position American technology to survive in a world where threats are increasingly moving into the digital space. In times of crisis, or cyberattacks from nefarious actors, distributing compute across individual nodes ensures continuity, whereas centralized systems collapse.
The way forward
So what’s the path forward?
We start by incentivizing distributed infrastructure, making it easier and more profitable to build beyond hyperscale facilities. We fund federal research and development for distributed computing to accelerate innovation in the public and private sectors. To host edge computing powered by local clean energy, we open up federal land and institutions. And finally, we streamline support for next-generation energy sources like advanced nuclear grids, so the future grid can match the volume of AI energy demand.
Through this approach, we reduce permit delays and unleash the latent value in our nation’s underused assets, from rural substations to decommissioned industrial zones. Our energy crisis cannot be solved with a single fix. But taken together, these steps serve as a resilient model for America to lead in AI development.
This shift does much more than fix our energy bottleneck—it reshapes access. Developers can build independently of Big Tech without begging for compute. These infrastructure policies would level the field for smaller players to build and deploy advanced AI models, decentralizing opportunity itself.
AI is set to shape every society and sector it touches. But ultimately, whoever controls the foundation will determine which values guide that outcome. We can let foreign powers consolidate that foundation, outstripping our capacities to build and entrenching centralization, surveillance, and control. Or we can leverage America’s edge and develop our infrastructure at the pace with which energy demands to guarantee resilience, transparency and freedom.
If the U.S. wants to lead in AI, we must act decisively. We cannot rely on legacy systems or lethargic bureaucracy. We don’t need more studies or more panels. If we want to define the future on our terms, we need to build, and we need to build now.
Let’s get to work.