


The episode examines whether Anthropic’s upcoming IPO could be a historic success or a sign of an AI bubble, focusing on its safety mission, financials, and the sustainability of massive AI infrastructure spending.
Erin Griffith, Silicon Valley reporter for The New York Times, specializing in tech and AI coverage.
Griffith covers Silicon Valley for NYT; she reports on AI companies and market dynamics.
Anthropic (founded by former OpenAI employees) is valued and anticipated because it’s a “pure play” large-language-model company and is first to IPO among major LLM firms. Its prospectus frames existential stakes around AI safety, but critics call the messaging exaggerated or distracting from present-day risks. Financially, headline losses are distorted by accounting on convertible notes; operating losses narrowed and the company is aiming for profitability, but it faces “knife’s edge” demand/compute commitments.
Anthropic’s refusal to give the Pentagon unrestricted access; concerns about fully autonomous weapons; reported cybersecurity contrast with OpenAI; comparisons to dot-com and housing-bubble overinvestment; hyperscalers’ AI spending and “neocloud” infrastructure debt; data-center community opposition.
Written by AI. May contain mistakes.
Discussion on Anthropic's rapid growth and IPO potential.
“And this week, they're one step closer to going public with what could be the biggest IPO of all time, up to$100 billion raised.”
Exploration of the high expectations surrounding Anthropic's IPO and its implications.
“And just based on your reporting, how much could Anthropics potentially raise and why is it so highly valued and anticipated?”
Insight into Anthropic's founders, their mission, and their departure from OpenAI.
“You know, if they come in below those really, really high expectations, it's going to be a different kind of signal.”
Analysis of Anthropic's revenue model and its implications for future growth.
“But Anthropic made some really interesting decisions early on that differentiated them from OpenAI.”
Discussion on Anthropic's financial health based on their IPO filing.
“They announced earlier this year that they plan to go public and then they kind of push the timeline back a little bit more.”
Exploration of sustainability concerns surrounding the AI market and infrastructure.
“Yeah, I mean, it's interesting because there are a lot lot of people who are kind of looking at this almost like a Roshar's test where it's like, you can see what you want to see.”
Discussion on how Anthropic's IPO reflects broader AI market trends and economic ties.
“I mean, you wrote this piece in the Times and the headline was, what's riding on Anthropics, market debut, nearly everything.”
Examination of Anthropic's unusual corporate structure as a public benefit corporation.
“What can the Luddites teach us about technology replacing human work?”
Explore the tension between shareholder value and long-term company goals.
“Because, you know, there is this problem with the markets and this idea of short termism where, like, you have to maximize everything for shareholder value each quarter or you'll be sued.”
Discuss the balancing act of safety in AI and the criticisms faced by Anthropic.
“where people are like, you're kind of over mystifying the machine god idea and playing into this idea that an AI system has consciousness and need to worry about model welfare.”
Analyze the comparisons between the AI market and historical economic bubbles.
“I just want to come back to the idea that this bubble is real.”
Investigate the challenges Anthropic faces regarding infrastructure and community response.
“So, yeah, I am closely watching those things.”

Get the gist, explore an idea or ask about something said.
On Front Burner, what are the key points of this episode?
Automatic transcript. May contain errors.
0:00Hey, I'm Farnoosh Tarabi, host of the So Money podcast. One thing I've learned from talking to entrepreneurs, your time is one of your most valuable resources. With Google Workspace with Gemini, Google's AI is built right into the premium versions of tools like Gmail, Google Docs, and Sheets. No new tools or extra logins, just AI to help you work smarter. Because less time spent on busy work means more time building the business you believe in. Get AI tools for work. Upgrade to Google Workspace with Gemini at goo.gle slash workspace with Gemini. This is a CBC Podcast.
0:39Erin Griffith: Hi, I'm Jamie Poisson.
0:49Erin Griffith: In just a few short years, Anthropics become one of the buzziest names in tech. In their own words, they're an AI company that puts safety at the frontier. And this week, they're one step closer to going public with what could be the biggest IPO of all time, up to$100 billion raised. There's a lot riding on it. The alarm bells of an AI bubble continue, and depending on how you look at it, Anthropic's 335-page prospectus doc for their IPO is either a promising pitch to investors, a terrifying warning about existential threats to humanity, or a look at the potentially unsustainable costs of keeping the AI train running.
1:28Erin Griffith: So to make sense of all of this, I'm talking to Erin Griffith. Erin covers Silicon Valley for The New York Times.
1:40Erin Griffith: Erin, thank you so much for coming on to FrontBurner. It's great to have you. Thanks for having me. Happy to be here. So Anthropics IPO could be one of the biggest ever, maybe even the biggest ever, right? Some projections had it surpassing SpaceX's$85.7 billion raised and their valuation of$1.8 trillion when they went public in June. And just based on your reporting, how much could Anthropics potentially raise and why is it so highly valued and anticipated? Yeah, I mean, and those numbers are kind of what outside analysts and outside investors and bankers are projecting. That's not necessarily what the company is putting out there.
2:21But these are really, really big numbers that we haven't seen before. And prior to SpaceX, I'd never seen anything close to it before. And for SpaceX is a 20 plus year old company, Anthropic has been around for only five years. So this is a really astounding moment. that's kind of almost like the cherry on top of the last four years of AI boom that we've seen all this money, all these literal trillions of dollars being poured into AI companies and AI infrastructure over that time. And so, yeah, as you mentioned, this is kind of turning into this almost reckoning or referendum or kind of bellwether moment for the AI boom.
2:59And it's almost a little bit bad for Anthropoc that these numbers have been out there because now with the markets becoming a little bit jittery as they have been over the last few weeks. You know, if they come in below those really, really high expectations, it's going to be a different kind of signal. And so, I mean, I think no matter what, it's going to be an enormous IPO. Will it be$2 trillion? Will it be$100 billion raised? You know, investors might be pulling back on that a little bit more, but it's still going to be enormous.
3:25Erin Griffith: I just want to better understand how this company that's just five years old, as you said, got here. These are former OpenAI employees. This is supposed to be the most successful IPO, I guess, in the history of capitalism. Tell me more about the founders, their mission, and how they got here. I mean, it is really unique and interesting company in that, you know, if you've read this prospectus or any of the stories about it, the letter that they kick off this financial document with is about, it's like almost a little existential. You know, it's about like humanity and consciousness and the most consequential technology ever built.
4:03It's kind of a heady letter from these founders, and it gives you a clue into what kind of people these are. They're not necessarily capitalist to the core, but as a byproduct of building this technology, they have built an incredible business. In its prospectus, Anthropic says AI could, among other game-changing acts, potentially cure diseases that have plagued humanity for millennia, including cancer. But it also says AI could be misused to design biological weapons or go rogue, which could be a global security threat of the first order. So as you mentioned, the founders of Anthropic peeled off from OpenAI to start the company because they believed that OpenAI wasn't concerned enough about the safety of the products that they were building.
4:49And it's a little bit ironic because OpenAI was created because the founders of that were concerned that Google DeepMind weren't worried enough about the safety risks of building AI. So that's kind of a theme at the heart of this industry. But Anthropic made some really interesting decisions early on that differentiated them from OpenAI. And one of them was that they were going to build software for businesses. They have Claude, which everyone kind of knows is a chatbot, but that's not really the focus of the company. what they make their money on is selling essentially tokens to businesses that use the software Claude and other pieces of software that they've built to be more efficient and essentially just do business better.
5:34OpenAI has kind of spent a lot of time messing around with image generators or video generators and all kinds of different things. And ChatGPT is kind of like their core product. And it wasn't until Anthropix coding product Claude Code really took off like gangbusters late last year, early this year, that OpenAI kind of saw that and said, OK, wait, we need to catch up here because this is a huge moneymaker. And, you know, that's partly what's driving Anthropik to the public markets first, is that they had this like just absolutely blockbuster product with Cloud Code and the growth is what we see in S1.
6:14Erin Griffith: OpenAI is expected to IPO next year. Yes. They announced earlier this year that they plan to go public and then they kind of push the timeline back a little bit more. So yeah, they're also rushing to the public markets. How do you think they feel about the fact that Anthropics now first? being first does mean you kind of get to define the terms and the valuation and the way that investors are kind of like analyzing your company you get to kind of like set the term similar to the way that on facebook went public um amid the kind of social media landscape they sort of you know said okay well you should look at our monthly active users and you know they defined all these terms in the same way that like a topic is going to kind of get to do when they go to go public first.
6:55They're the first like pure play, large language model to go public, to open their books and to show the world how the underlying business can work. So this is like the first time that Wall Street retail investors, the world is getting to kind of like dig in and see, does the math behind the business of this AI boom actually work?
7:16Erin Griffith: You know, just from what you can read in this S1 or this 335 page perspective, like, does the math work? Like Anthropic, I see, brought in$16.3 billion in revenue. They lost$80 billion in just the first half of 2026, though, although I guess$72 billion of that was a one-time charge. But, you know, I've also read that there's more to it than that. And just what are the takeaways from the filing on their financial health? Yeah, I mean, it's interesting because there are a lot lot of people who are kind of looking at this almost like a Roshar's test where it's like, you can see what you want to see.
7:55But, and so I've seen a lot, a range of takes on like, this is a terrible business and this is a great business. I've spoken with some analysts and, you know, investors who, you know, they, they think it's, it's a healthy and interesting and good business. They've been kind of almost pleasantly surprised by the fact that the company is shrinking its losses. And like you mentioned, the$80 billion net loss is a little bit deceiving because it's related to these convertible notes that they issued. And because the notes essentially went up in value, they have to write it down as a net loss. It's like an accounting quirk that we can kind of set aside.
8:32And if you look at their operating loss, so they lost$693 million in the first half, and that's narrowed from$2.1 billion the year prior. So they are finding ways to like make more money and on a path to profitability. However, they are spending enormous, enormous amounts of money building out the AI infrastructure to serve future customers. And so it's all really riding on their ability to correctly forecast basically how much demand there's going to be for their products. And Dario Amadei, the CEO and co-founder has talked about how they're kind of on a knife's edge when they're making these decisions on signing these deals.
9:14Some of these deals are like tens, hundreds of billions of dollars of commitments. Some of them they're not able to back out of. And so if they can't fill the demand, you know, they may have to try to resell them or they may be on the hook for them. And like, it could be, it could be really disastrous. Right. I could buy a trillion dollars of compute that starts at the end of 2027. And if my, if my revenue is not a trillion dollars, if it's even $800 billion. There's no force on Earth. There's no hedge on Earth that could stop me from going bankrupt if I buy that much compute. This kind of also speaks to a lot of the questions that have been bubbling up in the broader financial market around all these AI infrastructure companies and the quote-unquote neoclouds that basically repackage cloud computing for AI uses and resell it.
10:03Is this a sustainable market? It's very, very laden with debt, and there's been tons of money going into it. And there's a lot of questions around whether or not the infrastructure that's being built can get up and online and running fast enough to pay off the investments or to service the debt or to even just like start making some money back. And so that is, I think, where all of this is like, to use Dario's words, on a knife's edge.
10:30Erin Griffith: Yeah. I mean, you wrote this piece in the Times and the headline was, what's riding on Anthropics, market debut, nearly everything. I mean, is that what you're talking about there? Yeah. And it's a little bit hyperbolic just to kind of, but really, like, if you look at our economy over the last four years, it has sort of oriented itself all around AI. And like, you know, the top 10 S &P 500 stocks are worth more than a trillion together, and eight of them are tech companies. So the stock market is very much tied up in AI. There's been a trillion dollars in VC investment over the last four years into AI companies.
11:11The hyperscalers, Facebook, Google, Amazon, spending$740 billion this year alone. There is just so much money that has been poured into this industry on the expectation of continued growth. And that's why everybody is looking to Anthropic, this first pure play LLM company, to go public. That's the one that's driving the demand, creating the demand for the AI products for any signs that the demand might be slipping or slowing or not what we expected it to be.
11:58talking to entrepreneurs, your time is one of your most valuable resources. With Google Workspace with Gemini, Google's AI is built right into the premium versions of tools like Gmail, Google Docs, and Sheets. No new tools or extra logins, just AI to help you work smarter. Because less time spent on busy work means more time building the business you believe in. Get AI tools for work. Upgrade to Google Workspace with Gemini at goo.gle slash workspace with Gemini. Everyone's talking about AI, but to understand where it's going, you need to understand where it came from. What can the Luddites teach us about technology replacing human work?
12:39Can humanity learn from the nuclear age before creating something else it can't control? The future of AI didn't begin with ChatGPT. It began centuries ago. This is the global backstory, Superintelligence. intelligence. Five special episodes exploring the history behind the biggest story of our future. Listen to the global story wherever you get your BBC podcasts.
13:05Erin Griffith: One thing that was interesting to me about this filing is Anthropic operates as something called a public benefit corporation. I just want to get into that with you and what it means. The company alludes to it throughout the document. For example, they urged potential investors to quote, not think of your purchase in purely transactional terms and warned that some of their choices, quote, may run counter to investors' expectations because profit maximization will be balanced by the public good and business bets may seem strange or unorthodox in isolation. And I just, how unusual, how unusual is that?
13:49Erin Griffith: It's very unusual, right? It sounds like a weird kind of beam switch when someone's trying to tell you, like, buy my stock, but don't expect to make money on it. But at the same time, I can't help but imagine that every tech founder or maybe even every CEO in the world is looking at that and kind of thinking, like, I wish I had that. Because, you know, there is this problem with the markets and this idea of short termism where, like, you have to maximize everything for shareholder value each quarter or you'll be sued. And, you know, this is one of the reasons, if you were to argue in favor of founders maintaining control of their companies and not, you know, usually when you sell your shares, you have to give up the voting rights of those shares.
14:32but founders over the last 15 years, basically starting with Facebook, have figured out little workarounds and ways for them to basically take away any power from the shareholders and keep it for themselves and be able to appoint successors and be able to essentially veto any sort of shareholder proposals. And I think most people would argue this is bad governance and it's bad for the world when these extremely powerful and valuable companies are concentrating the power in the hands of like one or a handful of people with no checks and balances. But at the same time, the companies would argue, well, this is the way that we're able to make long-term bets that may lose us money in the short term, but will be better for the long-term sustainable health of the business.
15:17And then in Anthropik's case, they're arguing it's better for the survival of humanity, basically, because their S1 is basically telling us that if we don't allow their founders who are, according to them, uniquely capable of steering this technology in the right direction, then there could be, you know, catastrophic consequences for our very existence. So, you know, the stakes are high.
15:44Erin Griffith: Just this idea that this particular company is, you know, incredibly responsible and these founders are uniquely positioned to steer it in the correct direction. Like, does their track record so far back that up? Like, what do their critics, if they have them, I'm sure they do, what do they say? Yeah. On one hand, you have critics who would say that they are hysterical and maybe overplaying this idea of existential risk from AI. Like, obviously, there are a lot of harms and risks with this very powerful AI. And we're seeing a lot of this play out with a lot of the cybersecurity incidents that have happened in the last few months.
16:32But I think there are a lot of people who would argue that, like, this focus on ending humanity is like a distraction from the real tangible risks of today, where people are like, you're kind of over mystifying the machine god idea and playing into this idea that an AI system has consciousness and need to worry about model welfare. And like there's like a little bit of like a wackiness. There are other people who kind of point out to various hypocrisies where, you know, maybe they're saying that they're going to be self-righteous on certain things or they're going to stand up for themselves like they did against the Department of War.
17:04The company refusing to give the Pentagon unrestricted access to its AI technology. The company says it's concerned about the potential of fully autonomous weapons without human involvement. Late today, President Trump now banning Anthropic from all federal agencies. And they're also going to kind of capitulate when it comes to like regulation, because eventually it does come down to you have to do some real politic and do what it takes to keep your business alive. That is sort of one piece of it. On the flip side, their focus on safety has meant that they have not been involved in as serious of hacking incidents as, say, OpenAI has, which, you know, they had like incident after incident.
17:46And Anthropic has found a couple situations, similar cybersecurity situations as well, but there's not nearly as extreme as the ones that, you know, the main, the hugging face one that has gotten the most headlines. Right. Which was OpenAI, right? Yeah. Yes. Yeah. That was OpenAI. Also, to my knowledge, I haven't had as many situations where there's been harms from users using their software, like the sort of AI delusions or AI pushing users to commit suicide or helping users do bad things. So like, it's I'm not an expert on this, but it does seem to me like they haven't had as many like safety or harms as OpenAI.
18:25Although I think as the it's also probably a feature of the fact that it's not as much of a consumer product. And that's and they focused on more business use cases.
18:35Erin Griffith: And I guess like the big rhetorical question is like, is being the quote-unquote safest company of many in an industry moving at a breakneck speed still safe. Yeah, this kind of gets to the whole hypocrisy that people always are struggling to understand with the whole entire AI industry, which is like, if you think the software is so bad, why are you building it? And, you know, their answer is always like, it's going to be built. And so we need to be the ones who build it safely. And as that argument moves from these kind of small circles of people in this very small, tight-knit industry to a more mainstream audience, I think it's starting to get pressure tested a little bit.
19:20and it's not necessarily holding up to scrutiny in the way that they might have thought. And that's why you're seeing these calls from the leaders of the companies to, quote unquote, pace the frontier. And they need an outside group to help enforce a slowing down. So nothing really, really bad happens. We will slow down as much as necessary in order to make sure that every successive AI technology that we release is actually safe.
19:57Erin Griffith: I just want to come back to the idea that this bubble is real. Like the International Monetary Fund said in its latest Global Financial Stability Report that the global economy is increasingly at risk if there is an AI market correction. a growing number of analysts have drawn comparisons between what's happening here with AI and the housing bubble that led to the 2008 recession. Do you see those same parallels? You know, I mean, I think I've seen more parallels to, or at least in the world that I travel in, which is venture capital investors who are dumping money into this company, I hear a lot more comparisons to the dot-com bubble.
20:36And even, weirdly enough, they're favorable comparisons sometimes, where they're like, bubbles are good. Bubbles are what drives excitement for an industry. It brings capital and talent and momentum to something as it's being built from the ground up. And that's what a big technological revolution requires. And we saw this with the dot-com bubble. There was a lot of destruction along the way, a lot of stuff that made no sense, that completely collapsed. But proponents of this argument, which I'm not saying I take, but I've heard it a lot is that that was what gave us PayPal and eBay and Amazon and some of the kind of pillars of the early internet that emerged from this rebel because they were strong enough to like kind of make it out and they had good enough ideas.
21:25And at the same time, all the overinvestment was happening and, you know, laying the fiber in the cable that like got broadband connected to the entire country and that brought more people online and that, allowed the next wave of companies like the Facebook and the Web 2.0 companies to flourish. And so that's the argument that I've heard from VC investors. I think it's easy 25 years later to kind of say that because it's the scars from the fallout of that bubble collapse have faded.
21:57Erin Griffith: Right. If you were one of the people who lost their entire retirement savings in that crash. Exactly. And so I think there probably has, I'm imagining that a lot of the housing crisis comparison is happening on the debt side, which I've watched, but I haven't covered as closely. And I have seen a lot of nervousness and jerryness among investors there. And you're seeing that with the kind of rising yield prices for a lot of the debts that a lot of these companies are taking out, including SoftBank, which is one of the biggest investors in OpenAI. They're paying an extremely, extremely high interest rate on their latest loan.
22:40And they're saying, well, it's not that bad. And the core weave, one of the NeoCloud companies that is highly, highly tied into this ecosystem. They've had some taken some knocks lately from investors and, you know, one of their execs had a memo going around talking about how the markets just misunderstand the company. And when you see talk like that, you kind of have to worry that we're at the peak or at the precipice, the moment of where the crash is about to begin. So, yeah, I am closely watching those things.
23:14Erin Griffith: One thing I did want to ask you about, you know, we talked earlier about the kind of infrastructure that Anthropic is looking to build out, right? I mean, these data centers that they need to get built, the opposition to them, it really is only continuing to grow. It's become even a big issue in the midterms, right? If the success of companies like Anthropic is wholly dependent on it being able to have more of this infrastructure, could public outcry really throw a wrench in their plans here? Yeah, it has become a big issue. I feel like it's kind of faded from the headlines in the last couple of weeks with all the safety stuff taking front and center.
23:58But one thing I've also noticed during that time is that the companies are becoming a little bit more savvy about it. A big part of the outcry has been from people feeling like these companies are just barging in with their money, doing secret deals with their local politicians and just railroading over the communities with no input from them. And I think the companies are starting to wise up to that and introduce a little bit more transparency and realizing they're going to have to like pay what is essentially amounts to bribes to cities that need the money for their own infrastructure or schools or parks or whatever.
24:38And I think we're going to maybe find a middle ground there.
24:43Erin Griffith: I don't know. Maybe not. Erin, this is great. Thank you so much for this and just also for being such a trooper. you. You did this whole interview just with a cough. So thank you. Oh my God. Yeah. Sorry about that. And yeah, thanks for having me.
25:05Erin Griffith: All right. That's all for today. I'm Jamie Poisson. Thanks so much for listening. Talk to you tomorrow.
25:20For more CBC Podcasts, go to cbc.ca slash podcasts.
In the few years since its inception, Anthropic has become one of the buzziest companies in tech. In their own words, they’re an AI enterprise that “puts safety at the frontier”. And this week they’re one step closer to going public with one of the biggest initial public offerings of all time, with expectations that they could raise up to $100-billion.
There’s a lot riding on this. Anthropic’s market debut could prove that either the AI boom is here to stay or that the mountains of money poured into the AI race were a bad bet.
With Erin Griffith, tech and business reporter at the New York Times, we look at what Anthropic’s 335-page pitch to investors and the Securities and Exchange Commission tells us about the company’s structure, finances and whether they can keep the AI train running.
For transcripts of Front Burner, please visit: https://www.cbc.ca/radio/frontburner/transcripts