The American economy seems to be doing pretty well right now, at least on paper. Unemployment is low. Stocks are up from last year. Inflation is slowing. Things aren’t great for everyone, of course, but they could be a lot worse. President-elect Donald Trump hopes to upend that status quo in the new year.
On the campaign trail and since election night, Trump has vowed to impose significant tariffs on some of the nation’s largest trading partners, which will make things more expensive for the American consumer. Trump’s mass deportation plans could cause economic chaos if carried out at enough scale, particularly in the agriculture and construction sectors. His administration’s plans to disrupt and dismantle the federal bureaucracy could take a toll as well.
Trump tends to overpromise and under-deliver, so some or most of this may not happen. If his policies do lead to a recession, however, I think Americans will make an unsettling discovery: A significant portion of the American economy is essentially fake. Some of the nation’s wealthiest people are going to great lengths to keep it propped up, even to the point of swaying the presidential election to help preserve it. If anything, 2024 may be remembered as the year that the bubble took over.
Two separate but related industries exemplify what some have called the “bullshit economy.” The American Prospect’s David Dayen memorably wrote about it after Iowa Democrats paid a curiously well-connected tech company to develop an unnecessary app called Shadow to more speedily report the results of the 2020 caucus. When the app crashed on caucus night, bedlam followed, and it took longer to find out the results than if they’d done it the old-fashioned way.
“This is reflective of the rolling incompetence covered by confidence within the modern economy, especially when you sprinkle on the labor-saving promise of techtopia,” Dayen wrote. “When the bullshit economy fails, it robs people’s belief in the basic bargain of commerce, the idea that you get what you pay for, that companies operate in good faith to provide quality service.”
Two examples exemplify this trend. One of them is “crypto,” a term that now encompasses cryptocurrencies, blockchains, NFTs, and so on. Until recently it has been a grim time in the crypto world. In my 2022 retrospective column, I described crypto as the “boondoggle of the year,” partly because many of its highest-profile figures and companies had collapsed or been disgraced, with ruinous consequences for their customers and employees, and partly because “boondoggle” is a great word that I rarely get to use.
“Hopefully Americans exit this year with a healthy dose of wariness toward the crypto ‘industry’ and a deeper skepticism of get-rich-quick schemes,” I wrote at the time. How naïve I was. Some of crypto’s leading figures went to prison for fraud, tax evasion, and more. But the ones who didn’t are now stronger than ever. Not because crypto’s creators and evangelists have finally fashioned a productive or efficient use for it. Perish the thought. You are no closer to being able to buy a loaf of bread from your local grocery store with tether or ethereum or dogecoin or whatever the latest “coin” is than you were two years ago.
Instead, crypto’s survival has come through betraying its original ambitions. Crypto advocates have been correct about one thing over the years: The United States dollar, like nearly all modern currencies, is a fiat medium of exchange that is intrinsically vulnerable to inflation, deflation, and other forms of government manipulation. By unmooring itself from government control, the original, Great Recession–era creators of cryptocurrencies hoped to create an alternative to the dollar that would benefit ordinary people instead of banks and politicians.
The difference—what makes the dollar “real” and crypto “fake”—is that the dollar has an army, a navy, and courts that can render legally binding judgments upon creditors and debtors. Crypto’s wealthy backers have responded to this counterpoint in a simple, brute way—by simply buying Congress and the White House. The crypto industry spent at least $135 million on congressional candidates from both parties during the last election cycle, hoping to maximize their influence in Washington. It now enjoys support from Trump and the Republican Party, as well as a considerable number of Democrats.
What do they expect to get in return for these campaign donations? A more favorable regulatory and law enforcement environment, for one thing. Leading crypto figures know that all those pesky regulators and investigators in Washington—the Securities and Exchange Commission, the Federal Trade Commission, the Justice Department, and so on—could bring their party to an end. Trump has even pledged to make the U.S. a “Bitcoin superpower,” whatever that means. For now, it means installing pro-crypto figures atop the SEC and other key agencies.
In cryptocurrency, scams are so common that there is a special lingo surrounding them. Early adopters will create a coin, promote it, and then sell their supply at the top in what is known as a “rug-pull,” meaning that they pull the rug out from under the feet of their putative investors. Those left behind are known as “bag-holders.” The most recent one involved a young woman who became famous for a viral two-second joke about oral sex. For some reason, thousands of people bought a cryptocurrency that she promoted and brought its value up to $500 million—which then plummeted back to earth when the creators apparently cashed out.
The crypto industry’s greater ambition may be to get the federal government—and, by extension, all Americans—to be the ultimate bag-holder by directly buying cryptocurrencies with taxpayer dollars. Wyoming Senator Cynthia Lummis, one of the industry’s leading allies, introduced a bill this summer titled the Boosting Innovation, Technology, and Competitiveness through Optimized Investment Nationwide Act of 2024, or Bitcoin Act of 2024. (I’ve previously given my thoughts on acronym bills.)
If enacted, Lummis’s bill would require the Treasury Department to purchase 200,000 Bitcoins every year for five years after it is enacted. At current market prices, that would amount to roughly $10 trillion in real money. Those Bitcoins would then be placed in a “strategic Bitcoin reserve,” from which the federal government could not sell them for 20 years. Other Bitcoins already held by federal agencies would be transferred into this “reserve” as well. And then they would just … sit there.
This is a dream scenario for early crypto adopters and no one else. Forget trying to time the market’s peak before people inevitably catch on to crypto’s weaknesses or they fall into an unfavorable regulatory environment. They can simply unload all these unproductive, intangible, nonfungible, and inexplicably expensive assets onto the federal government of the United States and get real money for them in return. Lummis described it as “our Louisiana Purchase moment,” but that only makes sense if we’re the French.
While crypto has failed to achieve widespread use, AI is making extraordinary inroads into everyday life. It’s important to distinguish “generative AI,” which is the term I’ll use for the modern form, from actual artificial intelligence, which only appears in science fiction. Generative AI programs cannot replicate human intelligence. They lack creativity, imagination, self-doubt, and emotion. Branding triumphs over everything, however, and “artificial intelligence” probably appealed more to investors than “generative language learning models,” so now we are stuck with that term.
One of AI’s fundamental flaws is that it “hallucinates,” the term used by developers for when it gives wrong information based on its corpus of accurate information. Lawyers who use ChatGPT to research precedents for a lawsuit have been sanctioned for giving fake case citations in court. BBC News criticized Apple this month after the tech giant’s AI software incorrectly summarized its news articles to say that the UnitedHealthcare CEO murder suspect had shot himself.
There are some differences between these two industries. Crypto does not do anything or solve any problems; generative AI can at least do things, even if it currently or inherently does them poorly. There is genuine scientific and technological advancement behind AI software and the hardware that fuels it. It’s possible, if not probable, that generative large language models might have productive use cases in the near future. Even if Silicon Valley can’t create Data from Star Trek, it might be able to create, say, the universal translator or something extremely close to it.
The other problem is that people are mistaking existing AI products for actual artificial intelligence and using them in self-destructive ways. Even if one assumes AI will someday play a positive role in society, that day is not today. Major companies are stumbling over each other to launch AI-driven products and AI-connected services with little care for what consumers want or desire. When Apple added Apple Intelligence to its newest phones, most users reportedly found it largely useless.
AI’s top boosters aren’t aiming for something more humble, however. What they are promising is nothing less than a fundamental shift in the human experience. Marc Andreessen, the co-founder of Silicon Valley’s most influential venture capital fund, wrote a manifesto in 2023 that illustrates how they promote the new technology and dismiss its critics. Apparently not one to set low expectations, he titled it, “Why AI Will Save the World.”
Some of Andreessen’s points have merit. He is skeptical of the claims from some AI critics—and even a few of its enthusiasts—that AI could present some sort of extinction-level threat to humanity, dismissing them as either flawed misunderstandings of the technology or self-interested promotions by its boosters. (His ally Elon Musk, another AI hype enthusiast, is sometimes guilty of this.)
“The idea that it will at some point develop a mind of its own and decide that it has motivations that lead it to try to kill us is a superstitious handwave,” he wrote. “In short, AI doesn’t want, it doesn’t have goals, it doesn’t want to kill you, because it’s not alive. And AI is a machine—[it] is not going to come alive any more than your toaster will.”
Some of his other defenses of AI were less persuasive. In one section, he dismissed fears that AI will eliminate jobs and industries by noting that technology has always done that, and it has eventually led to new jobs and new industries. “We’ve been through two such technology-driven unemployment panic cycles in our recent past—the outsourcing panic of the 2000’s, and the automation panic of the 2010’s,” he wrote, noting that despite these fears “the world had more jobs at higher wages than ever in history” before Covid-19.
This argument likely plays better in Silicon Valley C-suites than, say, among steelworkers in Pennsylvania or automakers in Michigan who saw their livelihoods shipped off to foreign countries with weaker labor laws by corporate executives focused on maximizing shareholder value. The problem is not just that AI boosters hope to eliminate entire professions and industries, but that they are replacing them with inferior, hackneyed, hallucinating substitutes. UnitedHealthcare, for example, reportedly used it to deny medically necessary claims from elderly patients.
It also did not help Andreessen’s case that he undercut it himself almost immediately. After making a fairly reasonable argument for why technological innovation doesn’t necessarily reduce a society’s overall wealth, he let the mask slip a bit and invited his audience to “think of what it would mean for literally all existing human labor to be replaced by machines.”
It would mean a takeoff rate of economic productivity growth that would be absolutely stratospheric, far beyond any historical precedent. Prices of existing goods and services would drop across the board to virtually zero. Consumer welfare would skyrocket. Consumer spending power would skyrocket. New demand in the economy would explode. Entrepreneurs would create dizzying arrays of new industries, products, and services, and employ as many people and AI as they could as fast as possible to meet all the new demand.
Suppose AI once again replaces that labor? The cycle would repeat, driving consumer welfare, economic growth, and job and wage growth even higher. It would be a straight spiral up to a material utopia that neither Adam Smith or Karl Marx ever dared dream of. We should be so lucky.
Consider me somewhat skeptical that AI actually could do any of that. To be as specific as possible, the problem here isn’t that AI programs will render entire industries and forms of human endeavor obsolete and redundant. It’s that a lot of wealthy, powerful people think and want AI to do that, and are willing to replace flesh-and-blood Americans who have mortgages to pay and kids to raise with sloppy, hallucinatory chatbots that won’t take sick leave or request PTO.
Those who think this might be a net negative for American society are described as “doomers” by Andreessen and other AI investors for their Luddite resistance to their utopia. The solution, he argued, is for the government to make it as easy as possible for AI companies to immanentize the eschaton by shredding any troublesome regulations that could stand in their way. Any obstacles raised by broader societal concerns are dismissed out of hand.
“Big AI companies should be allowed to build AI as fast and aggressively as they can—but not allowed to achieve regulatory capture, not allowed to establish a government-protect[ed] cartel that is insulated from market competition due to incorrect claims of AI risk,” he wrote as part of his conclusions. “This will maximize the technological and societal payoff from the amazing capabilities of these companies, which are jewels of modern capitalism.”
And if you don’t give the oligarchs of Silicon Valley what they want, well, then you’re just helping Beijing. “To prevent the risk of China achieving global AI dominance, we should use the full power of our private sector, our scientific establishment, and our governments in concert to drive American and Western AI to absolute global dominance, including ultimately inside China itself,” Andreessen wrote. “We win, they lose.”
Another way to put this manifesto is that the Silicon Valley oligarchs who have invested billions of dollars in AI programs want to maximize their potential market share and reduce regulatory costs. They appear to think they are the first capitalists in history to argue that if they can make as much money as possible with no guardrails, it would be beneficial to society as a whole. They would be the first ones to be correct.
The problem with overpromising and under-delivering is that it catches up to you. Some Wall Street firms and Silicon Valley funds are already raising concerns about whether AI companies can achieve a return on the hundreds of billions they have spent. Sequoia Capital, one of the nation’s most prominent tech funds, noted last year that there is a growing gap between the amount of investment into hardware and software and the revenue that it generates.
Consider the following: For every $1 spent on a GPU, roughly $1 needs to be spent on energy costs to run the GPU in a data center. So if Nvidia sells $50B in run-rate GPU revenue by the end of the year (a conservative estimate based on analyst forecasts), that implies approximately $100B in data center expenditures. The end user of the GPU—for example, Starbucks, X, Tesla, Github Copilot or a new startup—needs to earn a margin too. Let’s assume they need to earn a 50% margin. This implies that for each year of current GPU CapEx, $200B of lifetime revenue would need to be generated by these GPUs to pay back the upfront capital investment.
Sequoia’s original version of that analysis was written in September 2023. When it revisited the question this summer, that number had risen to $600 billion of revenue needed just to pay off the initial investment. In the September version, it estimated that companies were only $125 billion off in annual revenue to meet that goal. Investment costs brought that number up to $500 billion when they reran the calculations this summer. These are roughly hewn figures, but they illustrate the scale of the problem: Way, way more money is being spent on AI than is being made on AI at the moment.
As a result, the valuations of some AI companies—and, thus, the wealth of their investors—are starting to look a little wonky. OpenAI, the top AI firm in town, recently announced an additional $6.6 billion in funding, bringing it up to a whopping $157 billion valuation. That would make it as valuable on paper as nine-tenths of the S&P 500 combined. Even though OpenAI had a net loss of $5 billion this year, it expects to make $100 billion in revenue—but not until 2029. Nvidia, the world’s largest chipmaker and a key supplier for AI companies, surpassed Apple as the world’s most valuable company last month at a $3.57 trillion valuation.
Again, there are important differences between crypto and AI, between a scam and excessive hype. But there are also some striking similarities in how the tech industry is selling them to the American people. Their top backers emphasize the importance of buying in early and lean heavily into the fear of missing out. They claim to be able to provide utopian or near-utopian outcomes, either by preventing the next Great Recession with a stateless, bankless currency or by creating a post-scarcity society of all-knowing AI servants. And they claim that the only way to achieve it is by dismantling any potential obstacles in their way.
With Trump’s reelection, 2024 represents their capture of the federal government and the chance to put all their plans into effect—unless the bubble bursts first.
