Last week, we had a podcast planning meeting, and my co-founder Rich asked, “When can we shut up about AI?” We all sighed. There are so many other things to talk and think about—including software things—but it feels like there isn’t much room for those subjects these days. Also last week, the Wall Street Journal published an article about Anthropic’s preparation for its IPO, and reported that Anthropic plans to tell investors that its TAM, or Total Addressable Market, is $30 trillion. That’s $30,000,000,000,000.00—a little less than the United States’s GDP. For contrast, all the residential real estate in the U.S. is worth about $55 trillion.

How can one think about that Anthropic number? I have a simple answer: Don’t bother. Just let it wash over you and live your life as if it never happened. Go ahead and erase this email. Have some ice cream if you’re able. Because that number, like so many others, has nothing to do with you. Nor with me. Even if you work in AI, it really shouldn’t bother you too much. It’s a number designed to excite large institutional investors.

Have you met many of them? They’re all basically huge money fans. It’s like your friend who is way too into Phish, but instead of watching hour-long jams on “Tweezer,” they enjoy watching money do things. These people would love to go to Vegas and watch a giant Bloomberg Terminal get projected on the walls of the Sphere. Anthropic’s number is only for them, in the same way that Phish’s music is only for Phish fans.

Still, even if it’s not your concern, you may, correctly, ask yourself: $30 trillion? Have the AI people lost their minds? I don’t think they have, not any more than normal. I do think they are completely divorced from the reality that most people think of as “reality.” But that’s different than losing your mind. They see a real logic behind the number, and they believe in it.

Now obviously TAM is not a real thing. It’s a speculative measure, and this is an investor pitch. So keep that in mind: Anthropic’s pre-IPO road show is the equivalent of a six-week residency at the Las Vegas Sphere for the world’s biggest money fans. The entire investment community is out there huffing nitrous and ingesting gold caps in the parking lot in anticipation of the greatest, most mind-altering show of their life, and they paid a lot for the hotel.

But let’s imagine it’s real. Where does all that money come from? Well, according to a recent widely-discussed post by Bill Gates, it will come from eliminating labor costs—from replacing humans. AI will, in his reckoning, take all the white collar jobs, then the manufacturing jobs. He thinks this process is already underway, and that we will need frameworks to manage humanity in the future. He even coined a term for it:

I like the phrase Human Reserved because it makes me think of nature reserves—places where we could put buildings and roads, but we choose not to because the loss would be too great.

We might set something aside as Human Reserved for economic reasons. For example, we may do it because allowing machines to take over a certain role will displace a large number of people who can’t easily change jobs. You can’t tell a 55-year-old who has worked in construction their whole career that they need to go work at an elder care facility and expect them to find it fulfilling.

He also calls for new taxes:

I believe we should tax AI tokens and robots. Right now, if you’re an employer and you hire someone, you pay payroll taxes on their earnings. But if you buy a robot, you can usually write it off right away as a business expense. The tax system nudges you toward replacing people with machines.

Reading the Bill Gates essay, it’s pretty easy to see how to get to $30 trillion: Take every job and every product, whether digital or physical, and imagine it can be made without human labor. (Someone would have to buy all this stuff, of course, but maybe robots will do that, too.) You’ll whisper, “Claude, buy me a few acres of land upstate, acquire the permits to build on it, get me some robots to 3D print a house, and more robots to deliver and install the house, fix all the problems, write me daily notes on progress, and then send a self-driving car to bring me there when it’s ready. Use my credit card.”

If you think the future could have that shape—and some number of powerful people really do—then an ungodly number of jobs go to job heaven. That is the labor force that sources materials, does architecture, expedites permits, drives trucks, builds houses, installs electricity and plumbing, and drives cars. Software ate the world, and now AI will eat software as an appetizer, then the global supply chain for dinner. For all that, $30 trillion seems like a bargain.

Gates thinks in vast systems. His last book was about climate change (which, speaking of big numbers, will incur a $38-trillion global yearly price tag by 2050, but they’re hoping AI will solve that, too). He’s also on a bit of an Epstein apology tour. And it’s worth noting—that thought experiment above, where you call a guy and get a house? Gates can do that today; it just requires a lot of humans. That’s his reality. The rest of us have Zillow.

We’re not gonna put the bots back in the box, so we should take steps to manage the future instead of letting it beat us up. There are three major axes of AI thinking: (1) “All hail the machine god!” where AI can do no wrong; (2) “Oops, all harms!” where it can do no good; and (3) “Love the player, regulate the game.” Like Gates, I’m in that third camp, and I, too, would like a thoughtful global governance framework in place to allow people access to great powers while limiting their risk. And a robot tax.

However, the world doesn’t work on my terms, or even on Bill Gates’ terms. A week ago, I hacked together a funny tool—basically a purpose-built RSS reader—to collect news from small newsrooms in the United States, including not-for-profits, small-press cooperatives, Black-owned newsrooms, and LGBTQIA+ newsrooms. There are a lot of them! Far more than I expected, especially if you mostly read the New York Times, like I do. 

The tool has only been fetching RSS feeds for a week, but in that time, it found more than 300 stories about data centers. These aren’t Facebook posts; these are articles where people went to a place and talked to people in order to serve their local community, whether in North Dakota, where they’re already starting to factor in the cost of de-commissioning the data center ($90 million), or in New Orleans, where they’re wondering what Meta actually plans to do in the community.

On a national level, there’s a great article by Jasmine Sun where she traveled around to see what people are saying about data centers:

Sarah—like many others—is worried that AI is a bubble. “They’re not making money, their business model’s failing, their product isn’t even that good,” she contended. In that world, could the AI companies still cover the cost of the grid expansion? Would they build the berms to hide the dirt piles? Would the buildings end up as stranded assets, abandoned like the dead factories, left as waste for her community to clean up? “It feels like supporting what happened right before the 2008 crash.”

How unpopular are they? According to Jeffrey M. Jones from Gallup, in May:

Seven in 10 Americans oppose constructing data centers for artificial intelligence in their local area, including nearly half, 48%, who are strongly opposed. Barely a quarter favor these projects, with 7% strongly in favor.

Reading through those data center stories, the same idea keeps leaping out: People feel that large, indifferent forces are taking advantage of them, and they don’t believe any promises the big platforms are making. They don’t trust the tech industry—which keeps explicitly saying it will take their jobs—so they’re doing the very American thing of loudly and annoyingly grinding sand into the gears until things start to slow down. One fire department in Oklahoma rejected a $250,000 donation from Google because the community around them didn’t want it, even though it was more than its annual budget. There are increasing numbers of lawsuits focused on slowing down data centers. These people won’t be motivated by a 3% robot tax and “Human Reserve.”

A lot of us live in a big-platform world and consume big-platform news, which covers other big platforms and federal systems. But if you look away from those platforms—places beside Silicon Valley, Washington, D.C., or New York City—then maybe the future of U.S. tech policy isn’t one big omnibus bill passed by a theoretically functional Congress in between lobbyist-funding canapé feedings, but rather thousands of little lawsuits filed against Apple, Meta, Google, Anthropic, SpaceX, Amazon, Oracle, and OpenAI. Is that the right way to do things? Of course not. But it’s what we’ve got. Like a lot of people, I’ve seen my ideal version of America get tossed around like a dog’s chew toy over the last few years, but I still retain faith in our national ability to be an enormous, chaotic pain in the ass when we want to be.

I have a 3D-printer spouse. Lots of little repair jobs around the house are starting to get fixed with tiny custom bits of plastic. I wonder, in a world where you can instruct a 3D printer in plain language, and where the printers keep getting smarter because code keeps getting cheaper, where that all ends up. 

There’s a different future vision where all of these wonderful new technologies are available in little factories at the end of the street, the AI lives on your laptop, and things get smaller and more local. The three big camps above have their AI fantasies: That it’ll go away forever, be regulated intelligently, or will achieve AGI. Mine is that all this pandemonium and sense of abandonment leads to people building tighter, more resilient communities. Regardless, we finally got what we all said we wanted: A national conversation on the role of technology in our communities. It just happens to be about data centers.

So that will play out, awkwardly, with politicians wading in further (Bernie and AOC are way in front of it), part of the midterms and then the presidential race. There will be moratoriums everywhere, and tech companies will be shocked—shocked!—to find out just how little people trust them. (Meanwhile, China will look on, bemused, and bring another terawatt of solar power online.)

And in the meantime, we’re going to see ever bigger AI numbers. Now that we’ve crossed the speculative $30-trillion mark, Elon Musk will need to point to a $100-trillion TAM for whatever he whips up next. OpenAI will need to get in the mix. Before long, we’ll be talking about quadrillions of dollars. Then we’ll really be cooking. And every time the number gets bigger, people will start protesting data centers even more.