A lot of monopoly-related news, as usual. The Senate agreed to fund the government as long as no one touches health care corruption, the Supreme Court is likely to strip Trump of a key part of his tariff authority, and Disney and Google feud over which monopolist gets to control sports programming. Also, Lina Khan content now goes way more viral than that of Barack Obama.

But this week, the most important story are the results of the elections Tuesday, which I suspect reflect a deep dissatisfaction not just with Trump, but with American economic strategy since the mid-2000s. Here’s pollster G. Elliot Morris describing the key dynamic: “Among voters who said the economy was their top issue, partisan trust on the economy moved 93 points toward Democrats between 2024 and this year’s New Jersey and Virginia elections.” In addition, this week, Trump’s disapproval collapsed, moving from an average of negative 7.5% for most of his first year to negative 13% just this week.

These shifts, while large, make sense, and actually track what happened to Joe Biden. Indeed, Trump’s politics in some ways mirror those of Biden. For instance, during the Biden era, Democrats obsessed over niche identity questions and ignored the real economy, throwing a few crumbs towards an industrial policy, which is to say subsidies for chips, batteries, and solar panels, and a bit of antitrust. Wall Street and big tech did very well, and prices were high. The result was that Republicans over-performed.

This time, it’s Republicans obsessing over niche identity questions and ignoring the real economy, while throwing slightly different crumbs towards an industrial policy, which is to say tariffs plus subsidies for chips and critical minerals. Wall Street and big tech are doing very well, and prices are high. This time, the Democrats over-performed.

Taking a step back, voters have chosen change over incumbents in virtually every election since 2006. And it’s not a coincidence that this ‘throw the bums out’ era was organized around a particularly financier-friendly form of statecraft, one based on the government fostering excessively high returns on capital. The methods were monopolization, a low labor share of income, private equity extraction, and Wall Street-led growth through bubbles in subprime housing or data center build-outs.

Of course, Donald Trump’s economic strategy is not just straight continuity with what came before, so it’s important to look at specifically what he did, and why voters are saying no. The short story is that Trump’s first term had a focus on using trade and immigration policy to raise wages. Trump’s second term is about helping to direct most resources in the American economy towards the build-out of AI data centers. And that has raised wages, but mostly for the people pictured below at his inauguration.

This picture wasn’t some isolated incident. The day after his inauguration, Trump gathered with more tech CEOs to announce Stargate, a $500 billion build-out of data centers by OpenAI, Softbank and Oracle. During the press conference for Stargate, Trump pledged emergency declarations to build more data centers, and OpenAI CEO Sam Altman discussed how AI would cure diseases and grow the economy. Symbolically, featuring big tech on day one and day two of an administration was an important statement.

Indeed, while the press has focused rightfully on many social questions, Trump’s economic statecraft really does reflect the policy preference of big tech CEOs and Wall Street. Trump’s most important business advisor is Jensen Huang, the CEO of AI chip firm Nvidia. From January to May, Tesla CEO Elon Musk ran DOGE, which orchestrated widespread changes in government workflow. Then in June, Trump signed the Big Beautiful Bill, which was a bunch of tax cuts for the wealthy, and especially for big tech firms.

That bill almost included a White House-backed ban on states doing regulation of AI, but the moratorium was only pulled after a furious fight within Congress.

Wall Street has at this point fused with big tech into one giant interest group inside the administration, focused on undermining populism and working people. For instance, Trump could have chosen populist trade expert Robert Lighthizer, his first term trade advisor for the Treasury Secretary role. But instead he went with billionaire Scott Bessent, a Soros-employed former currency trader who has focused on bailing out hedge funds in Argentina and supporting Wall Street’s goal of higher stock prices. Bessent wants big tech to do well, because the top seven stocks - the “Mag 7” - now represent roughly 40% of the stock market.

The same month as the Big Beautiful bill passed, Trump came out with his administration’s “AI Action Plan,” with the goal of achieving “global dominance in artificial intelligence.” This plan is a serious statement of intent. Important advisors like David Sacks had orchestrated the firing of the Librarian of Congress due to disputes over copyright restrictions on the development of AI. Trump has continued to follow through, with AI executive order after AI executive order. AI lobbyists like Kyrsten Sinema are threatening local communities to allow data centers, saying if they don’t the Federal government will preempt their authority anyway.

In September, Trump hosted a task force on AI with top CEOs, where each threw out absurdly large investment pledges of data center buildout.

American Central Planning

What’s odd about this strategy is that this data center-based strategy is not actually profitable. It’s a money sink right now, a bet on the future. And yet, Trump’s AI-first strategy has a logic. According to Paul Kedrosky, investments in AI data centers now outpace the telecom boom of the dot com era, and are starting to resemble railroads in the 1880s. Without the data center buildout, the U.S. might be in a recession.

Moreover, for Trump, if data centers require lots of energy, Americans can supply it by drilling for oil and gas. If data centers require lots of capital, Wall Street can supply it by attracting capital from all over the world. If data centers keep investors buying stocks, then U.S. stock markets get to keep going up. And if data centers are so massive that they demand a substantial amount of national resources, well, the U.S. does need GDP growth that this bubble can supply.

Every morning on CNBC, Jim Cramer discusses all of the companies dependent on data center build-outs, and it’s increasingly more of the old-line industrial firms who supply various component parts, like turbines or specialty chemicals to data centers, as well as financial firms who are supplying capital and are dependent on an increasing stock market. Big tech firms are going to invest four hundred billion in data centers next year, and that’s pretty much what building trade unions are focused on. The American stock market, and increasingly the economy, depend on the AI data center buildout.

Given the lack of profits, what’s the point of the investments? Well, investors, and Trump, are betting on a story. These big tech firms, goes the narrative, will be able to extract further profits from the economy as a result of hundreds of billions of data center investment in AI, which will transform the nature of work. Even as Trump tries to explain how good AI is going to be for America, various big business leaders are explaining to their investors how AI is going to allow them to get rid of workers and raise returns on capital. The inconsistencies at the heart of these stories are obvious, and apparent to everyone.

The Logic of a Bubble: Number Go Up

And what of the technology itself? Is generative AI actually useful? Yes. It has a lot of different applications, across multiple industries. Many of the apps in your phone already use forms of machine learning, as do industries as varied as medicine and entertainment.

But whether generative AI is useful is besides the point. AI data center build outs are now American industrial policy. Five big tech monopolies have grabbed an increasing share of profits in the economy, and they are engaged in a form of central planning to force investment in bloated construction projects. Wall Street is following along with capital, and firms like McKinsey and corporate influencers are forcing every corporation to impose, top-down, AI on everyone, whether it makes sense or not.

And there are several problems with planning an economy around developing AI data centers on behalf of tech monopolists. One is that that you will develop the wrong form of the technology. As Azeem Azhar has noted, Chinese AI models are as good and cheaper than American models, because Silicon Valley and big tech prioritize compute heavy technology, and China does not. Nvidia, Amazon, Microsoft, and Google sell the equivalent of gasoline, they want gas-guzzlers on the road, not efficient cars that get the job done.

The more important question is as follows. Is generative AI what we should actually be dedicating our time, capital, energy, and water to creating as a society? The promise of AI boosters is that they will cure diseases, transform the American economy, make us all more productive, destroy jobs, whatever. But it’s not that it will do what Americans actually want, which is make life more affordable. On Twitter, I noted how companies are laying people off with AI as an excuse, but none of them seem to be using AI to lower prices. That tweet went viral.

If Americans saw clear benefits from AI, like lower prices at the supermarket, they would embrace it. But they aren’t. And frankly, the last major technological “advances” - the smartphone and social media - have been bad for most people. We now expect that technology deployed by oligarchs will be used as digital leashes, not tools of liberation.

The truth is that Wall Street and big tech firms are all investing in a known bubble, in the hopes that it will somehow work out, but if it doesn’t, well the government will backstop them. Already, Altman is preemptively asking for a bailout, and the Trump administration has had to deny that one is in the offing.

And while I’ve pointed to Trump, that’s mostly because he’s in charge. Elites on the Democratic side see the AI industrial policy the same way. The Biden administration bet on AI data center build-outs. This week, Democratic Presidential hopeful and Michigan Governor Gretchen Whitmer just pointed to an OpenAI Stargate data center investment pledge, bragging that she helped win “the largest economic project in Michigan history.” Another possible 2028 candidate, Democrat Josh Shapiro of Pennsylvania, said his state is ‘all in’ on data center development.

These comments are all part of an ideology. Since the 1980s, America has done national economic development through financial bubbles, which is to say, by allowing Wall Street rather than democratic institutions to organize where we allocate capital. This form of statecraft is a result of a theory put out by the founder of private equity, former Nixon Treasury Secretary William Simon, who invented the idea of a “capital shortage” in the 1970s. According to Simon, American governance didn’t guarantee sufficient returns to capital, so there wasn’t enough investment. His framework involved cutting capital gains taxes, deregulation to juice returns, and government guarantees and bailouts to reduce risk for investors. And his vision is still dominant.

The savings and loan real estate bubble, the dot com bubble, the housing bubble, and now the AI bubble are all functions of this odd form of economic statecraft. A few months ago, I discussed this dynamic as the “Number Go Up” rule.

Increasing the capitalization of the stock market at all costs has a number of implications about how our society works. Anything hindering short-term increases in profits, whether that’s higher wages, more factory investment, ending tax loopholes, rules to block being able to unsubscribe from Planet Fitness, transitioning to different energy sources, addressing an opioid crisis, or breaking Chinese control over the necessities of life, gets pushed out of the way. American life spans have plateaued or stagnated since 2015, which you’d think would be of concern. But it’s just not. There’s also something that distinguishes standard financial capitalism from ‘number go up.’ Financial capitalism implies risk. But in our era, the government guarantees financial returns with subsidies, regulations and bailouts. It’s a form of statecraft.

There’s an institutional design part of this vision. Pushing the stock market requires relatively few government levers, just the Federal Reserve and some tax experts and fancy lawyers. AI data centers are an easy way to channel capital under this model, since what it really takes is a few guys in Silicon Valley cutting huge checks for something that is easily understandable. By contrast, populist form of government requires huge numbers of people and institutions to oversee the messy work of running a complex economy with lots of small businesses doing different things, and training high-wage workers to make things.

In other words, the reason we are engaged in this massive wasteful data center buildout has nothing to do with technology, it’s simply that bubble-driven growth run by a few dominant firms and backed by Wall Street is the only form of national development strategy we know how to do.

And Yet…

As I noted above, the election last Tuesday is just the latest rebellion against Number Go Up statecraft. And this time, it’s anger at AI buildouts. Indeed, almost every week, as Heatmap points out, “from arid Tucson, Arizona, to the suburban sprawl of the D.C. area, Americans are protesting, rejecting, restricting, or banning new data center development.” Candidates won office entirely on rejecting data centers.

Even among conservatives in the non-big tech business world, there’s anger. Here’s the libertarian founder of logistics firm Freight Waves, Craig Fuller, making the point.

And yet, while there’s a lot of skepticism and anger about data center build-out, the build-out is accelerating. Opposition is similar to the failed attempt to regulate big tech.

Why Isn’t There a Coherent Opposition to AI-Centric Development?

We are spending $400 billion or more next year on data center build-out, no one seriously doubts that there are far better ways to spend that. For instance, America needs to be able to feed ourselves, clothe ourselves, and make our own medicine, none of which we are able to do now. We need to have a society that is great for families and human flourishing. We need support for kids, for small businesses, and for education. We need cheaper housing, better cars, food processing capacity, more farms, less expensive electricity, a merchant marine, and nicer airports.

However, all of these require the ability to actually plan something, execute on that plan, and channel capital through institutions like banks into functional markets, without middlemen eating up profits and erecting bureaucratic barriers. We can rebuild that capacity, but we have to start with the courage to imagine a different society. That’s in its initial stages. Biden did some of it, with chips, batteries, and solar panels, and the concept of “industrial policy.” So there are outlines, it doesn’t all need to be done from scratch. But what Biden did was limited.

And you can see how limited the liberal imagination is with the disinterest in important institutions like the Federal Reserve or banking regulation, which are seen as technocratic and boring. There must be an awareness of how excess returns to capital are driving dysfunction. And populists must build out different ways to channel capital. So far, that isn’t happening, even if Americans understand data centers are toxic.

Can Trump Move Away from AI-First?

One of Trump’s strengths is that he is flexible, and what we have started to see is a signal that he might move away from AI dominance as his approach to economic development. His political director is saying so. But the most obvious signpost was his post on TruthSocial that he wants an investigation into meatpackers, since prices for beef are high and ranchers who sell cattle are very angry.

But I’m not convinced it’s possible to pivot at this point. His Treasury Secretary, Scott Bessent, and his Commerce Secretary, Howard Lutnick, have stocked the administration with Wall Street and big tech friendly operatives. Meanwhile, his Antitrust Division chief, Gail Slater, hasn’t brought a single merger challenge. Even if Trump did want to return to populism, he’d have to fire Bessent and bring back someone like Lighthizer, his trade advisor in his first term. And Senate Republicans just rejected his demand to end the filibuster, so he’d have to work with Democrats. It’s just hard to see him doing what is necessary to shift gears.

The most likely path over the next few years is that Trump slowly bleeds out, as Biden did, leaving behind a shattered wreck of a party and movement. Whether the opposition party can put together a coherent critique of the AI data center build out strategy, and propose an actual national development strategy, well that’s the question.

And now, the rest of the round-up after the paywall. An ugly end to the shutdown as the health care industry triumphs, Trump’s tariff strategy looks like it’s going to get whacked by the Supreme Court, Lina Khan content is way more viral than Barack Obama content among Democrats, an unexpected and positive roadblock for Hollywood consolidation emerges, and libertarian Argentine President Javier Milei, doing the Trump YMCA dance in a tuxedo, somehow becomes a critical figure in the re-emergence of antitrust in America.

Read on for more...