AI is no longer an emerging technology. It is here, and it is becoming the basis of modern civilization. Just as electricity transformed the twentieth century and the Internet transformed 21, AI is remodeling how we work, govern and live. Soon, each important institution, from hospitals to the army, will integrate the AI in its central operations, increasing the infrastructure bets that support it.
Despite this demand, our infrastructure does not keep the pace. In 2024, US data centers used ~ 200 Terawatt-Horas of electricity, enough to feed Thailand for a year. The same estimate argues that by 2028, it is expected that the use of range between 165 and 326 Terawatt-Hora annually, sufficient to feed 22% of US households. The workloads of AIs are pushing energy and calculate the systems far beyond their limits, creating an exponential demand that lets our electricity grid left behind while fighting even incremently.
This mismatch is more than a technical problem. As AI’s demand increases, these bottlenecks in the national energy supply and calculation access will reduce development in each sector, which limits its transformative potential.
The United States is leading, for now. But we are in a sprint, and China is gaining ground. Its R1 Depseek model rivals the American first level models. Deepseek’s success shows that speed, scale and efficiency can radically change the balance of the global power of AI. The impulse of China is well financed, coordinated and strategic. If Deepseek is an indication of China’s impulse, we are far behind them.
It will not matter who leads the algorithms if the US continues to treat infrastructure as a late occurrence, because we are on the way to losing the platform war. The future of AI should be based on freedom, transparency and trust, not surveillance and control. That is the advantage of the United States, and for that we must prioritize the energy crisis that is creating.
In this context, mass and centralized data centers are obsolete. They are rigid, expensive and confined to a geographical location. Worse, they create individual failure points. If an energy grid falls or overheats, a complete segment of the country immerses itself in a technological dark era.
On the contrary, decentralized systems release our potential, allowing US innovation to scale with agility. The smallest computing groups can work near sources of localized renewable energy, such as solar, wind or geothermal energy, or take advantage of underutilized computing power in homes, campus and communities. Decentralized systems also place American technology better to survive in a world where threats are increasingly moved to digital space. In times of crisis, or cyber attacks of nefarious actors, the distribution of computing through individual nodes ensures continuity, while centralized systems collapse.
The way to follow
So what is the way forward?
We begin by encouraging distributed infrastructure, which makes it easier and more profitable to build beyond hyperscale facilities. We finance federal research and development for distributed computer science to accelerate innovation in public and private sectors. To organize the edge of edge promoted by local clean energy, we open federal lands and institutions. And finally, we simplify support for next -generation energy sources such as advanced nuclear networks, so the future network can match the volume of AI energy demand.
Through this approach, we reduce permissions delays and unleash the latent value in the infrautilized assets of our nation, from rural substations to dismantled industrial areas. Our energy crisis cannot be solved with a single solution. But together, these steps serve as a resistant model for the United States to lead the development of AI.
This change makes much more than fix our energy bottleneck: reorganizes access. Developers can build independently of Big Tech without begging for calculation. These infrastructure policies would level the field for smaller players to build and display advanced AI models, decentralizing opportunities themselves.
AI is ready to shape all the societies and sectors that it touches. But ultimately, who controls the base will determine which values will guide that result. We can let foreign powers consolidate that base, overcoming our capabilities to build and strengthen centralization, surveillance and control. Or we can take advantage of the advantage of the United States and develop our infrastructure to the rhythm with which energy requires guaranteeing resilience, transparency and freedom.
If the United States wants to lead in AI, we must act decisively. We cannot trust inherited systems or lethargic bureaucracy. We do not need more studies or more panels. If we want to define the future in our terms, we need to build and need to build now.
Let’s work.
