We will build the future?

Realizing the benefits of AI requires reviving our ability to build.

Tim Fist and Arnab Datta shared an unsettling fact:

Globally, the power required by AI data centers could grow by more than 130 GW by 2030, whereas American power generation is forecasted to grow by only 30 GW, much of it unavailable or unusable for AI data centers.

The implications of this mismatch are profound. Let’s assume (as I think we’re safe to given that Anthropic just earned a $61.5 billion valuation) that AI development is here to stay and, in fact, will likely accelerate. The lack of power generation available in the US leaves firms with a simple decision—seek that power elsewhere.

In the abstract, this seems like a perfectly fine outcome. Specialization has long been the name of the game when it comes to our global economy. Why develop something domestically if you can find it elsewhere at a lower price?

In practice, reliance on any external source for a critical input renders our political and economic well-being more fragile. As we’ve sadly seen in Europe, critical infrastructure has become a tool of exploitation and retaliation. Since Russia’s invasion of Ukraine, EU households have seen their energy bills surge by more than 30 percent. Europe could have avoided undue reliance on Russia by doubling down on building out domestic capacity. Instead, countries like Germany were forced to tap into coal and other fossil fuels.

Back to the power necessary to sustain the inevitable AI wave—the obvious solution is to scale up domestic power sources. That simple solution, though, has become needlessly and frustratingly complex. Again, here’s Fist and Datta:

To meet immediate power needs, many companies are considering on-site natural gas plants. Investing in natural gas plants to power AI data centers is a straightforward business decision, especially given the rapid deployment potential of combined-cycle gas turbines, which can be built in as little as three years. However, while natural gas is likely to remain a key part of the overall energy mix, it faces uncertainties that could encourage investment in other technologies. Supply chain shortages, existing decarbonization commitments, regulatory uncertainty, and stranded asset risk all make the decision less straightforward than it might initially appear[.]

These barriers to scaling up natural gas are replicated across other sources of energy. Whether its building a new nuclear plant or creating a new wind turbine farm, a mountain of regulatory sludge (though often well-intended sludge) has slowed our ability to build.

Meanwhile, other countries have managed to address these hurdles head on. Here’s a revealing depiction of how the US has become a laggard in energy generation:

A failure to remedy our inability to build will expose our economy and, by extension, our national security to increased risks. Our largest AI labs will invest in opportunities abroad to power the models of the future. That possibility should give us all pause and increase consideration of the proposals outlined by Fist and Datta.

The duo first call for an all-of-the-above strategy to power generation. Their thorough report demonstrates that solar, wind, nor geothermal alone will suffice to meet our energy needs. Instead, we need to rapidly invest in those and other sources of energy.

They also point out that legal hurdles to building more, sooner can be lowered through existing laws such as the Defense Production Act. A product of the Korean War, the DPA affords the president broad authority to remove certain procedural checks on massive projects, such as laying transmission lines.

Arguably their most important finding is found between the lines of their report: sufficient domestic energy production will only come about if it becomes a national political priority. This is no small task. Regulatory barriers to infrastructure projects isn’t the stuff of Main Street conversations. Even on the Hill, other issues seem to have distracted Congress from seriously debating these matters. That’s not to say that sparking such interest is impossible but simply to show that it’s an uphill battle.

I really enjoyed talking with Tim and Arnab and encourage you to give the podcast a listen when you’re doing the dishes tonight. Don’t get me wrong, I don’t think a podcast will initiate the public focus that’s required for progress on this front. However, it’s a start. Once you’ve learned just how far we have to go to build the future, you’ll find that it comes to dominate your thoughts. Yes, we have plenty of problems that deserve our immediate attention. AND, we need to divert some of our political capital and attention to the very real, practical matter of building our future.