Introduction: Toward a New Developmentalist Politics
We need higher quality discourse around AI and data centers.
AI is the most obvious vector of growth in our economy. A 2025 Harvard analysis found the vast majority of GDP growth in the first half of that year came from data center capex.
Workers, particularly in the construction trades, are cleaning up financially on the strength of these investments. Most biotech and biopharma companies are using AI in some capacity. Companies like Palantir and Anduril are using their AI capabilities to help the U.S. win — even better, prevent — wars and secure its borders.
Noticing these positive developments doesn’t require anyone to buy in to the most bullish predictions about record GDP growth, cures for cancer and Alzheimers, or eventual post-scarcity.
The Great Stagnation looks to be ending, and there’s a lot to be excited about. But not everyone is excited.
While the Trump Administration and states like Mississippi and Texas have pursued sound policies around AI, I began to worry sometime last year that the conservative movement was turning against growth and technological progress.
In fact, economic growth has become something of a bogeyman for much of the online right. There’s this notion that everything people find enviable about mid-20th century lifestyles was sacrificed at the altar of GDP. There are quite a few a memes about this, and they get tens of thousands of likes every time someone reposts them.
The reality is, the mid-20th century was a much more economically dynamic time overall, and you can see this — of all places — in the GDP numbers.
A 2019 analysis by the San Francisco Fed found that slow growth had become “the new normal” for the American economy, driven mostly by an aging workforce. The analysis suggested that we would need massive productivity gains to approximate the kind of growth our grandparents enjoyed.
You can just as clearly implicate low growth in the escalating affordability crisis. Overly rigid land use policies, preemptive environmental impact reviews, and even effective moratoria on entire technologies like nuclear energy have limited the scope of economic progress. In many areas, we’re no longer allowed to do more with less, and high prices are the market’s way of telling us that.
The United States of America, the frontier empire which first eyed the west and later the moon, is badly in need of a frontier. AI has come not a moment too soon.
It’s in some ways the perfect technology for our specific ailments. By enhancing productivity, it can help us restore growth and tame inflation.
It’s also more dependent on physical infrastructure than digital technologies traditionally have been, opening up opportunities for workers who didn’t enjoy as much direct positive exposure to previous revolutions in computing and software. (I had a piece on this last week in American Intelligence, one of my favorite publications.)
Unfortunately, the physical side of AI means it’s also vulnerable to the same forces that have suppressed growth in other parts of the world of atoms, to which software was largely immune.
Last month, Maine passed a statewide ban on new data centers. At the federal level, Bernie Sanders and AOC have proposed legislation to ban them. In many places, local opponents have succeeded in shutting down projects.
An analysis from Data Center Watch found $64 billion in planned data center spending has been blocked or delayed. Some of these projects may have been legitimately bad fits for their planned locations, but there’s enormous danger in allowing a million veto points to kill projects that are strongly in the public interest.
And this is assuming all opposition to American AI is well-meaning and public-spirited. There are good reasons to suspect our adversaries are at a minimum pleased with the opposition. To give a comically on-the-nose example, the dean of the Beijing Institute of AI Safety and Governance attended Sen. Sanders’ recent AI event.
Now is a time for choosing. America will take this last, best chance to escape stagnation or it will slip into the same cycle of degrowth and national decline that the rest of the western world has gotten trapped in.
With that in mind, a couple of months ago, I started a new initiative: the American Compute Project. The goal is to promote policy and messaging that will pave the way for a vital and reindustrialized America, with computing infrastructure as the obvious starting point.
Right now, AmCompute’s output is mostly X threads:
We’ve been able to build a small, well-positioned community around these ideas and are working on scaling our output. This newsletter is the first step.
We have maybe a year and a half to shape the conversation on AI and data centers before the next wave of primaries. The tenor of that conversation will in turn shape policy for the following two, four, six years and beyond, amid a possible inflection point in human history.
Welcome to Computing Infrastructure. I’d love if you’d subscribe, share, and encourage your likeminded friends to do the same.