


Anthropic’s IPO filing and what it reveals about profitability, future chip/data-center costs, and AI safety; plus a separate segment on Alaska midterm politics centered on fishing, and a science explainer on redefining the SI second.
Miles Krupa (Economist technology correspondent) interviews about Anthropic’s IPO prospectus. Aaron Braun (Economist West Coast correspondent) reports from Kodiak Island on the Alaska Senate race.
Anthropic’s Q2 revenue was $11.6B with >$900M operating profit, but future capital expenditures and long-term data-center leases will likely crimp near-term profits. The company is shifting from relying on cloud chip procurement (Google/Microsoft/Amazon) to buying chips directly (e.g., Broadcom) and leasing/installing in its own data centers. Anthropic is a public benefit corporation, aiming to balance safety and profits despite “catastrophic risk” concerns.
“public benefit corporation” framing; “USDA office of seafood” debate in Alaska; NOAA weather/stock data reliability fears from federal cuts; cesium-to-new-atom clocks (e.g., lutetium) for redefining the second.
Written by AI. May contain mistakes.
Exploring Anthropic's official IPO filing and market significance.
“and the need to change how a second is actually defined.”
A deep dive into Anthropic's financials and growth trajectory.
“So what is it we can now see in the financials here?”
Anthropic's transition to direct chip procurement and data center leasing.
“Well, one question is why Anthropic is doing this IPO now?”
Discussion on Anthropic's approach to safety and long-term sustainability.
“Or how much will it show us that all of this is the bubble that many fear?”
Summarizing key takeaways about Anthropic's IPO prospects.
“I think Anthropic has a really positive story to tell.”
Exploring the significance of fishing in Alaska's political landscape.
“It's about 9.30 in the morning, my first morning on Kodiak Island in Alaska, and I just arrived at Kodiak High School.”
Explore the dynamics and stakes of the Alaskan political landscape.
“That she even has a chance is a reminder of how Alaska's politics are remarkably fluid, even in such a partisan era.”
Learn about the proposed changes to the definition of a second and its implications.
“I choose these words because for nearly 60 years, the whole world's definition of precisely how long a second is is dependent on atoms of the element cesium.”

Get the gist, explore an idea or ask about something said.
On The Intelligence from The Economist, what are the key points of this episode?
Automatic transcript. May contain errors.
0:00With the Spark Cash Plus card from Capital One, you earn unlimited 2 % cash back on every purchase. And you get big purchasing power, so your business can spend more and earn more. Stephen, Brandon, and Bruno, the business owners of SandCloud, reinvested their 2 % cash back to help build the company's retail presence. Capital One, what's in your wallet? Find out more at CapitalOne.com slash Spark Cash Plus. Terms apply. This Economist podcast is sponsored by Apollo.io. In a fast-moving business environment, finding the right people to connect with takes time. For sales teams, prospecting, outreach, and follow-up can mean juggling multiple tools and workflows.
0:41Apollo.io brings these activities together in one platform, helping sales teams find the right buyers and spend less time on scattered, manual tasks and more time building relationships. Apollo.io. Sign up for free at Apollo.io.
0:58The Economist.
1:05Hello and welcome to The Intelligence from The Economist. I'm Jason Palmer.
1:14Today on the show, we're off to Alaska as our Midterm Battlegrounds series continues and the need to change how a second is actually defined.
1:29First up, though.
1:38It was almost as if you could hear the thump of 300-odd pages hitting the internet. Yesterday, Anthropic at last revealed its official filing to go public. It would be the first frontier AI company to do an IPO. And that has an entire industry watching and nibbling its nails. Like many AI watchers who have been waiting, our technology correspondent, Miles Krupa, has been sifting through the prospectus. Anthropics growth has been nothing short of astonishing. It's basically gone from zero to potentially two trillion in the span of five years. However, as this perspective shows, there are still a lot of long-term questions hanging over the company, especially around just how much it will need to spend on chips and other computing resources to fulfill its vision of creating exponentially powerful AI.
2:38So what is it we can now see in the financials here? The financials are pretty extraordinary. area. Anthropic showed full financials through the first half of this year. And just in the second quarter alone, it made$11.6 billion in revenue. And when you compare that to the first quarter of 2025, when it released the Cloud Code AI product, that's a 30-fold growth in revenue from that first quarter of 2025 to the second quarter of 2026. So that's pretty remarkable on its own. The other remarkable thing is that unlike a lot of private companies of its size and age, it actually managed to turn an operating profit in the second quarter of more than$900 million.
3:28So those two things are making investors really excited about Anthropik's potential as a public company. But now Anthropik wants to go public. it has put out a prospectus. What do we know from that that we didn't know before? It gave us a lot more detail into how Anthropic is thinking about the cost side of the equation going forward. Anthropic has generated a nice operating profit pretty early in its life as a company. However, it's laying out a future in which it will have to spend a lot more money up front to get the kind of computing resources it needs to continue staying on the frontier of AI and continue being able to serve its products to its users.
4:17So what we learned is that even Anthropic is kind of guiding investors that the profit that you see today may not represent the profit that we're able to make in the future, in part because their capital expenditures are going to go up in the short term. And they're also entering a lot of very long-term obligations to rent data centers, for instance. Well, one question is why Anthropic is doing this IPO now? Because Anthropic will clearly need a lot of capital in the future to purchase the computing resources it needs. One reason for it to go public now is to try to get a solid credit rating. So It's not that unlisted companies never get a credit rating or never get a good credit rating, but the more of a track record Anthropic can build as a profitable public company, the quicker it will be able to get an investment-grade credit rating, which opens up a much huger market for its debt and ultimately lowers its costs.
5:21So from what you can see now, what is it that Anthropic is going to have to do not for its IPO, but after to realize its potential? So Anthropic has relied very heavily on procuring chips through cloud providers like Google, Microsoft, and Amazon, the big three. But what's clear in the prospectus is that Anthropic realizes it needs to take these matters more into its own hands. So it started buying chips directly from vendors like Broadcom and installing them in its own data centers that it's leasing. So it's basically moving from this model where it's renting from the big cloud companies to building more of its own computing resources, which is a really capital intensive endeavor.
6:12and something that Anthropic is trying to sort of carefully lay out for investors, trying to show them that this shift might crimp profits in the near term, but may long term be better for the company's profitability and its control over its own destiny. And to your mind, how much of it is a sort of test case for the wider AI ecosystem? How will other companies learn from it? Or how much will it show us that all of this is the bubble that many fear? I think on the bubble question, Anthropic right now is dispelling or quieting a lot of the fears of the AI bubble because Anthropic is showing that you can build a profitable business around AI, at least in the short term.
6:58What it will say for other AI companies, I think a lot of that will depend on how the IPO goes. But I think we can say just from the early reception to the paperwork that it released this week, that people are eager to see profits from AI companies. And so for a company like OpenAI that has been spending a lot of money on data centers, perhaps more and earlier than Anthropic, it may be a bit of a wake-up call to get a bit closer to profitability than it might otherwise. And what about AI safety? Did you pick up any clues about how Anthropic is thinking about that? I mean, as with any other company going public these days, Anthropic has a pretty lengthy risk factors segment of the prospectus.
7:51And as you might imagine, there is a lot of talk in there about the risks that its models can go off the rails. There are multiple mentions of potential catastrophic risk, the difficulties in even insuring against these kind of risks. But then also I think Anthropic's paperwork is raising a bunch of interesting questions about how the company plans to balance safety and profits in the long term. Anthropic is what is known as a public benefit corporation, meaning that it has some leeway to not just maximize profit for shareholders, but to also consider other factors in the way it runs the company.
8:33And that's a very deliberate choice by Anthropic because they do want to try to prioritize AI safety as well as profits. So all told, in terms of the ethos of all of this, Anthropic is, if you like, taking the sort of high ground. It is your AI safety company. It is now your IPO transparency company. It has its eye on safety and a sort of sustainability of the model into the future. This is all fairly positive if you're gunning to put your money into the company. Totally. I think Anthropic has a really positive story to tell. There are still some unresolved questions about how exactly it plans to balance safety and profits going forward.
9:12We haven't seen a company try to balance so many different factors as a public benefit corporation at this scale. And I do think there are long-term questions about just how much these massive investments in computing resources will weigh on anthropics profits. But for now, the picture looks pretty rosy. Miles, thanks very much for your time. Thank you.
9:53Here's a feeling you might know. You come out of back-to-back meetings, open your inbox, and there are a hundred new messages. Somewhere in there is the thing that actually matters, and you're not totally sure what it is. That low-grade worry that something's slipping through? I've tried AI for this. It helped, but it lived in another tab. Too much time spent copying, pasting, and trying to write the perfect prompt, and too little time doing the work I actually needed to get done. Then I found Superhuman Go from the makers of Grammarly. It works where you already work. Reading a dense thread? Highlight it.
10:27Ask Go for the key points and you're already up to speed. Not sure what you owe after those meetings? Ask what was decided and get clear action items. Before you hit send, Go flags a number or name that doesn't add up. It's right there when you need it, on the side of your browser, without switching apps or losing your place. And you're always the one in control. Try Superhuman Go from the makers of Grammarly. Find out more at superhuman.com.
11:17at CapitalOne.com slash SparkCash Plus. Terms apply.
11:29It's about 9.30 in the morning, my first morning on Kodiak Island in Alaska, and I just arrived at Kodiak High School. The big congressional fisheries debate is happening in a few hours at the high school's auditorium. But before I even stepped foot inside, I noticed that the sidewalk outside the high school is absolutely covered in chalk paintings that the kids must have done. So there's normal kid stuff like hopscotch. But then I get over towards the door and I see a kind of chalk mural that says fish for the future. And then you get a little bit closer, and in bold print it says we live on fish.
12:16Few things matter more to Alaskans than fishing. Aaron Braun is our West Coast correspondent. That much was clear to me as I began reporting on Kodiak Island, on Alaska's southern coast.
12:32This year, fishing also dominates the state's politics. The issue is front and center in the Senate race between Dan Sullivan, the Republican incumbent, and Mary Peltola, a former Democratic congresswoman. This race is one of just a handful that will determine whether Democrats win control of the Senate in this year's midterm elections. So fishing might seem provincial to some voters elsewhere who are fighting about Medicare for All or immigration, but Peltola seems to be betting that she can overcome the Democratic Party's toxic brand in rural places by ignoring national issues in favor of local ones, like fish.
13:22Alaska is an unusual state with unusual politics. Fewer than 740 ,000 people live in a territory twice the size of Texas. — Donald Trump is the winner in Alaska, picking up three more electoral votes. — President Donald Trump carried the state by 13 points in 2024. In fact, he was just in the state in October campaigning for Dan Sullivan. — That was a lot of flying. In fact, they were thinking about canceling. They said, sir, you're very busy. This is very tough. I said, you think I'd ever cancel? I don't cancel for Alaska. So Mary Peltola has to win over a lot of Trump voters in order to win.
14:04That she even has a chance is a reminder of how Alaska's politics are remarkably fluid, even in such a partisan era. For example, the ruling majority in the statehouse is a tripartisan coalition of Democrats, Republicans, and nonpartisans. That's unheard of in other states. Alaskans, Democrat and Republican, just kept telling me that party identity matters less here. I've been fishing in Alaska since I was six years old. I became a permit holder when I turned 12. I'm the only candidate on this stage who's an Alaskan fisherman. Fish is in my blood, earned over generations on the water. When Mary Peltola took the debate stage in Kodiak, she looked the part.
14:54She was wearing rubber boots and a sweatshirt that said, proud coasty mom. Two of her sons are in the Coast Guard, and that resonates here in Kodiak. The island is home to the largest Coast Guard base in the Pacific. Alaskans know the difference between a senator who fights for fish and one who fishes for donations. Peltola is a native Alaskan who's proven that she can win statewide. In 2022, she beat Sarah Palin, a former Republican governor and vice presidential candidate, to become Alaska's lone voice in the House of Representatives. Thank you to the chamber. I love this community. I've learned so much from this community.
15:33Sullivan seemed comfy enough on stage. Kodiak residents told me that he visits their island fairly often. In fact, I think he was on my flight on the way over. As the campaign has progressed, it has become more contentious. During the debate, the candidates often took shots at one another. Sullivan argued several times that Peltola was ineffective when she was serving in the House. Here's the bottom line. Our fishermen deserve relentless focus and action, not just talk. During my opponent's first term in Congress, she did not introduce one bill dealing with fish. And as for Peltola, instead of acknowledging some of Sullivan's accomplishments, like pushing the Trump administration to create an office of seafood within the Department of Agriculture, She argued that he hasn't gone far enough to help Alaskans.
16:22We should be fighting for our fishermen. And yes, we do need a USDA office of seafood, but we can't celebrate just yet. We need to make sure that that USDA office of seafood is fighting for wild stocks, Alaskan stocks, not farmed fish. Not the Dan Sullivan USDA office of farmed fish. For the uninitiated, connecting your opponent to farmed fish is a major diss in Alaskan politics.
16:54The debate just ended. I'm waiting if I can steal a minute with Mary Paltola. Hey, welcome to Alaska. Erin, nice to meet you. I am writing about the ways that federal funding cuts and job cuts are affecting Alaskans. Just wondering specifically how that's affecting fishermen in your state. Well, the biggest gap that we are seeing is not enough updates on weather and what our weather is. When Peltola does talk about national issues like the war in Iran, federal cuts or energy policy, she does so strategically. She doesn't insult President Trump directly because a lot of Alaskans still support him.
17:34But she does attack some of his policies. For example, Alaskans have been particularly affected by the Department of Government Efficiencies, or DOGEs, dismantling of the federal bureaucracy. The percentage of federal employees is higher in Alaska than anywhere else. A recent report found that Alaska lost nearly 5 % of its federal workforce in 2025, a higher share than any other state other than Maryland and New Mexico. Officials and fishermen rely on data from the National Oceanic and Atmospheric Administration, or NOAA, to set seasonal fishing quotas, track fish stocks, and monitor the weather.
18:14Scientists are worried about that data getting less reliable as a result of those cuts. During the debate, every time Paltola brought up Trump administration policies that have been painful for Alaskans, Sullivan would grimace and look down. There isn't much more he can do. Were you a fan of either of the candidates coming in? Yes, definitely a big fan of Mary Peltola. What do you like about Ms. Peltola? Oh, well, she's, I think, honest. Her main thing is she's just quite nice. Yeah. A number of people I spoke with mentioned how approachable Peltola is. But a recent investigation from The New York Times suggested that she can be a chaotic or even cruel boss.
18:59Her campaign denies any pattern of abuse and points out that many of the anecdotes brought up by disgruntled former staffers occurred in the months following her husband's death, the time when she was grieving. But the race is extremely close, and because many Alaskans live in remote places, it could take days or even weeks to learn the result. It is my last morning on Kodiak Island and I have wandered down to the harbor just to see the island's fishing fleet. It's really overcast and cloudy today, rainy, pretty gloomy and I'm just staring at boats aplenty. I can see the names of a lot of them.
19:43We have the Atlas, real fun, R-E-E-L, Ventress. In such a big state with many far-flung communities, the fishermen on Kodiak Island were really pleased to see their candidates show up to ask for their votes. Now they just have to cast them.
20:09Our correspondents have been zigzagging across America throughout the campaign season. This week, they'll keep bringing you the details of some of the most illuminating races. Tomorrow, we're going to Georgia.
20:50The End cesium atoms. I choose these words because for nearly 60 years, the whole world's definition of precisely how long a second is is dependent on atoms of the element cesium. But now metrologists, the patient folks who dedicate their lives to matters of measurement, are having second thoughts. Earlier this month, the world's metrologists agreed on a fundamental change. David Adam writes about science for The Economist. The General Conference on Weights and Measures says it wants to redefine the second, which is the official SI unit of time. Well, let's start with how the second is defined as it stands.
21:34Well, we've come a long way since we used the rotation of the Earth to define time. For the second, we now use cesium-133, a particular isotope. And we use that because when you expose this to microwave radiation at a particular frequency, The cesium atoms, they switch between energy states, a bit like a tick in a clock. And so you can use them as a clock. You can tune the frequency of the radiation using the atoms so it'll keep it in time, precisely on target. The SI second is defined as exactly 9 ,192 ,631 ,770 cycles of that radiation. It's used to set international atomic time, which is used to coordinate everything from financial transactions to satellite navigation.
22:27That atomic time is set by something called the International Bureau of Weights and Measures. Why decide to change how the second is defined? Atomic time, it's so important that we get it right, that we use hundreds of very, very advanced clocks. And we take an average of that, and that's actually used to set time. So the SI unit is to define time. To set time, we use lots of different clocks. Now, over the years, some of those clocks have got much better than the cesium clock that we use to define the second, which was invented in the 50s. We're now into clocks that run on atoms or ions from more exotic elements, strontium, ytterbium, aluminium.
23:09And these clocks can be about 100 times more accurate than the best cesium clock. It creates a bit of an awkward, embarrassing problem for the metrologists, because to measure the frequency of one of these very accurate clocks in hertz, you've got to measure their performance based on the SI second. The SI second is still defined by cesium, which is less accurate. So in a way, it's a bit like trying to measure the width of a hair using a ruler with divisions that are wider than the hair. It won't affect us in our everyday life. It's not going to change how long it takes to boil an egg or whether the train will turn up on schedule.
23:47But it's a wrinkle in the system of time. And so how to iron out the wrinkle? To do that, they need to release the definition of the second from cesium using one of these more accurate modern clocks. Now these clocks are usually developed by national labs, some independent researchers at universities, but there's lots of different options. One example would be a clock based on lutetium. It has a tick accurate to about one part in 10 to the 19, which is an extremely big number. One with 19 zeros after it. There are other advanced clocks, not quite as good as Lutetium, but still 10 to the 18 range, which is about a hundred times better than even the best cesium clock.
24:31That doesn't make Lutetium the best candidate to define the second. There's quite an involved process to go through. And one of the things you need to do is to show that these clocks work reliably at different places in the world. And the Letitium clock is as big as a room, so they can't put it anywhere else. So it's probably going to be one of the other clocks that are quite as good as Letitium, but it's still much better than cesium. One of the issues is not science, it's politics. Lots of countries have national laboratories that are developing their own versions of the clock. And I think it's fair to say that getting the second defined by your clock is something of a feather in a metrologist's cap.
25:08So the people who are going to decide which version to use are mostly from some of these national laboratories. Essentially, you're going to have to be voting on perhaps someone else's design. David, thanks very much for joining us and explaining all that. Thank you.
25:38That's all for this episode of The Intelligence. We'll see you back here tomorrow.
26:01With no fees or minimums on checking accounts, It's no wonder the Capital One bank guy is so passionate about banking with Capital One. If he were here, he wouldn't just tell you about no fees or minimums. He'd also talk about how most Capital One cafes are open seven days a week to assist with your banking needs. Yep, even on weekends, it's pretty much all he talks about. In a good way. What's in your wallet? Terms apply. See CapitalOne.com slash bank. Capital One N.A. Member FDIC.
It would be the first public listing from a frontier-AI firm—and perhaps the largest IPO of all time. We comb through the filing to understand how the company is thinking about itself and its industry. Our American-midterms series continues in the battleground state of Alaska. And scientists consider how to redefine the second.
Guests and host:
Topics covered:
Listen to what matters most, from global politics and business to science and technology—subscribe to The Economist.
Hosted on Acast. See acast.com/privacy for more information.