As some of the biggest companies in AI commit billions of dollars to building new data centers, the need for electrical power has emerged as a major constraint. So much so that the U.S. Department of Energy (DoE), recognizing the adoption of AI as a national strategic priority, has called for ideas for clearing the bottleneck. The DoE calls this initiative “Speed to Power.”

The nation clearly needs ever greater amounts of electrical generating capacity to meet our AI data center expansion goals. The good news is that nearer term there’s an opportunity to make better use of the existing national power supply – including the growing capacity of green energy – by reframing the problem. Instead of focusing only on how and where to generate and transmit the gigawatts, let’s consider where to locate the gigabits.

A fast-emerging next-generation data transmission technology is bringing new geographic flexibility to where data centers are built. That technology is hollow-core fiber (HCF).

In terms of higher speed and lower “latency” – disruptive data transmission delays – HCF is vastly superior to the optical fiber that has been the industry standard for the past 40 years. The best HCF is roughly 47% faster than conventional fiber and reduces latency by 33%. Thus, data centers can operate further away without sacrificing the latency their customers depend upon. That means people no longer have to worry about data centers being in their backyards and companies aren't pigeonholed into building data centers next to their existing offices.

Together, these HCF attributes can enable a data center to be placed closer to the best current sources of electricity. What’s more, the data centers can also be built farther from cities or suburbs where zoning restrictions can cause long delays in the approval process for construction of the new centers. Think of HCF as a substitute for a transmission line; one which is cheaper, connects the data center to the grid faster, and addresses many “not in my back yard” (NIMBY) issues that plague data centers and other infrastructure development.

Strategic recommendations

In responding to the DoE’s call for ideas, we made the case for HCF to be included among the strategic tools that can enable the United States to stay ahead in the global contest for AI supremacy.

Our main recommendations:

• Recognize and include connectivity technologies such as HCF and its data center geographic flexibility as strategic tools in the “Speed to Power” mission.
• Use mapping and analytics from the DoE and National Labs to identify zones of underutilized power infrastructure to accelerate data center/AI development and improve grid-asset use.
• Establish criteria in DOE funding and program frameworks to support the use of HCF – as well as the domestic manufacturing of HCF and related optical technology through supply-chain resilience initiatives that align with national security and strategy infrastructure goals.
• Encourage pilot projects of HCF-connected data centers. Such pilots could be even more future-facing by including quantum computing applications, which also require high-speed, low-latency data transmission.

The power of fresh thinking

AI has a potentially transformative role to play in our nation’s business, economic, and societal future. But the adoption of this powerful technology requires wise public policy. That’s why it’s great that the DoE is soliciting the industry’s best thinking about how to speed the construction of the data centers that will be the engines of the nation’s AI economy.

HCF technology isn’t the only solution to opening the data center bottlenecks, of course. But the nation’s AI strategy demands fresh thinking about moving the gigabits closer to the gigawatts.