Monologue: AI Isn't Too Big to Fail

3 Apr 2026 · 18 min · 8 chapters

Ask about this episode

Ask anything about it. ChatGPT or Claude reads this page and answers with the times it was said.

Connect VO and ask about every podcast you hear, including the moments you saved. Add to ChatGPT · Add to Claude

In short

Ed Zitron argues that the “subprime AI crisis” is real—AI businesses rely on venture capital subsidies and user token burn rates that won’t survive once pricing and rate limits tighten. He compares it to the 2008 subprime mortgage bubble: lax underwriting and “easy money” created a mirage of viability. He cites examples like Perplexity burning $30–$100+ of tokens on a $20 plan, Anthropic historically allowing $8–$13.5 token burn per $1 subscription revenue, and Harvey (AI for lawyers) raising $200M at an $11B valuation with only ~$190M ARR. He then claims AI isn’t “too big to fail” like AIG was: AIG’s collapse would have frozen commercial paper funding and insurance, while an OpenAI/Anthropic collapse would hurt markets but wouldn’t collapse core US credit/insurance systems. No guests.

Written by AI. May contain mistakes. Listen to the episode to check what was said.

Chapters

Tap a time to open that second in VO

Upcoming Guest Announcement

2:53 to 3:04

Teasing next week's guest and future topics.

Subprime AI Crisis Explained

3:04 to 4:00

Ed discusses the parallels between the AI industry and the subprime mortgage crisis.

“In reality, the value of housing was massively overinflated by the lax standards of a mortgage industry incentivized to sign as many people as possible thanks to a lack of regulation and easily available funding.”

The Illusion of Viable Businesses

4:00 to 4:52

Analyzing how AI startups are funded and their unsustainable business models.

“You might know that, but I've been encouraged by Robert and Sophie to be obvious with things.”

Examples of Unsustainable Funding

4:52 to 6:20

Detailed examples of AI companies relying heavily on venture capital.

“a business that only works, and I put that in quotation marks, as long as venture capital continues to fund it.”

Consequences of Economic Failures

6:20 to 8:05

Discussing potential failures of AI companies if venture capital dries up.

“Much like subprime loans allowed borrowers to get mortgages they had no hope of paying, hype cycles create the illusion of viable businesses that cannot and will never survive without the subsidies.”

Too Big to Fail Discussion

8:05 to 9:19

Exploring the concept of 'too big to fail' in the context of AI companies.

“As AI companies made up more than 50 % of venture capital investments in 2025, I actually don't know how any of them make it.”

The Role of Bailouts

12:01 to 14:00

Discussing the specifics of financial bailouts during crises.

“Oh, and also the PDCF and TSLF finance systems that kept them going too.”

The Economics of AI and the Financial Crisis

14:00 to 21:19

Explore the distinctions between the AI industry's risks and the systemic failures seen during the financial crisis.

“market via the primary dealer credit and term securities lending facilities that provided as much as$100 billion to banks and financial institutions a day.”
Hear the part that matters, and keep it.Open this episode in VO. Double tap your headphones to save a moment as you listen.
Get VO free

Transcript

Automatic transcript. May contain errors.

0:00This is an iHeart Podcast. Guaranteed human. Support for the show comes from Public, the investing platform for those who take it seriously. On Public, you can build a multi-asset portfolio of stocks, bonds, options, crypto, and now generated assets, which allow you to turn any idea into an investable index with AI. It all starts with your prompt. From renewable energy companies with high free cash flow to semiconductor suppliers growing revenue over 20 % year over year, you can literally type any prompt and put the AI to work. It screens thousands of stocks, builds a one-of-a-kind index, and lets you backtest it against the S &P 500.

0:34Then you can invest in a few clicks. Generated assets are like ETFs with infinite possibilities, completely customizable and based on your thesis, not someone else's. Go to public.com slash podcast and earn an uncapped 1 % bonus when you transfer your portfolio. That's public.com slash podcast. Paid for by Public Investing. Brokered services by Open to the Public Investing, Inc., member FINRA, and SIPC. Advisory services by Public Advisors, LLC, SEC Registered Advisor. Generated assets is an interactive analysis tool. Output is for informational purposes only and is not an investment recommendation or advice.

1:02Complete disclosures available at public.com slash disclosures. Shake it up with Vital Proteins Collagen and Protein Shake. It's a high-quality, ready-to-drink shake with 30 grams of protein and 10 grams of collagen to support healthy hair, skin, nails, bones, and joints. With zero grams of added sugar, no artificial sweeteners, and absolutely no carrageenan, it's a clean, delicious way to fuel your day. so you don't just age gracefully, you age powerfully. Vital Proteins. Stay vital. Learn more at vitalproteins.com. Apple Vacations, where your story starts. Spring is the perfect time to bring everyone together.

1:43Spring into savings with Apple Vacations, America's favorite vacation company. Book by April 9th and take up to$400 off international and Hawaii vacations or up to$125 off domestic getaways. Plus, kids stay free at select resorts when you travel through December 15th. From Cancun to Jamaica to Punta Cana and beyond. Reconnect in paradise. Plan your getaway at applevacations.com or contact your trusted travel advisor. Apple Vacations, where your story starts. And now, The Laundry Hero, brought to you by Grand Appliance. Oh, really? Okay, I'm on it. Uh-oh, what's up? Yeah, laundry's piled up and my washer just died.

2:19That's a bummer. Check out grandappliance.com. My brother got a great deal on a Maytag washer. Grand Appliance? My parents shop there. Mine too. Whole family. I'm on the site. Wow,$5.97 with next day delivery? Problem solved. Shop Grand Appliance. Appliance experts since 1930. Call Zone Media. Hello and welcome to this week's Better Offline monologue. I'm your host, Ed Zitron.

2:52Better Offline. that's right it's your second damn monologue this week but next week we're gonna have an awesome guest economist paul kodroski and no doubt some sort of news will break that will make next week's monologue even spicier than this one and man is this spicy anyway earlier this week i put out a free newsletter about the subprime ai crisis my ongoing theory that as ai companies strive to try and make their rotten economics work they'll have to start cranking up the price and making rate limits worse and generally trying to move things around to make these products anything close to profitable, but they won't even get close.

3:28Now, take a little history trip. That's what I'm calling it. When the subprime crisis happened, the subprime mortgage crisis, of course, millions of people built their lives around the idea that easy money would always be available and that anyone could get a mortgage and that housing would only ever increase in value. In reality, the value of housing was massively overinflated by the lax standards of a mortgage industry incentivized to sign as many people as possible thanks to a lack of regulation and easily available funding. The value of housing, and indeed the larger housing and construction boom, was a mirage.

4:01In reality, housing wasn't worth anywhere near what it was being sold for, and the massive demand for housing was only possible with unlimited resources and lacks, well, underwriting, which is the process of evaluating whether someone can get a house. You might know that, but I've been encouraged by Robert and Sophie to be obvious with things. Those buying houses they couldn't afford with adjustable-rate mortgages either didn't understand the terms or believed members of the media and government officials that suggested housing prices would never decrease and that one could easily refinance the mortgage in question.

4:32You know, kind of like saying things like, you know, AI's always getting more efficient, the value's always going up, and venture capital will always invest, and that this is the new hyper-growth era. Similarly, AI startups' products are all subsidized by venture capital and must, in literally every case, allow users to burn tokens far in excess of their subscription fees, a business that only works, and I put that in quotation marks, as long as venture capital continues to fund it. And when I say that, I'm being quite literal. If you go and use perplexity, you're comfortably able to, even on a$20 plan, burn$30,$40,$50,$100 worth of tokens within a calendar month.

5:08Anthropic allowed you, until very recently, to burn anywhere between$8 to$13.5 dollars per dollar of subscription revenue. While from the outside these may seem like these are functional businesses with paying users, without the hype cycle justifying the endless capital, these businesses wouldn't be possible, let alone viable in any way, shape, or form. And indeed, they wouldn't have any customers, my evidence being, if they could get customers by charging their actual rates and offering a non-subsidized product, they'd have them, and they would have had them from the beginning. Let me give you an example.

5:41Harvey is an AI tool for lawyers that just raised$200 million at an astonishing$11 billion valuation, all while having an equally astonishingly small$190 million in ARR or$15.8 million a month. It raised another$160 million in December 2025 after raising$300 million in June 2025 after raising$300 million in February 2025. Where's the fucking money, Harvey. Where are you putting it? Harvey. Mr. Harvey has been very unfair to the venture capitalists. Actually, they're fueling it. I can't even say that. Remove even one of those venture capital rounds and Harvey coughs up blood and dies. Much like subprime loans allowed borrowers to get mortgages they had no hope of paying, hype cycles create the illusion of viable businesses that cannot and will never survive without the subsidies.

6:30The same goes for companies like OpenAI and Anthropic, both of whom created priority processing tiers for their enterprise customers in the middle of 2025, and the latter of which, just as I discussed in my last monologue, added peak rate limits from 5 a.m. to 11 p.m. Pacific time, you know, just the entire day, and that was after they created weekly limits late last year. Their customers are the subprime borrowers too. They built workflows around using these products that may or may not be possible with new rate limits. Think about it. If your whole business, the whole reason you use this software subscription is to do tasks and suddenly the amount of tasks you can do is limited, is this really, is anything tenable anymore?

7:10Especially if you're one of those people who can't actually code and are using this to vibe code. I'm not really sure this works. But in the case of the enterprise customers using priority processing, their costs massively spiked, which is why Cursor and Replit and several other AI startups suddenly made their products worse in the middle of 2025, adding their own rate limits and changing their pricing. In reality, none of this ever made sense. None of this was actually possible outside of endless resources. Now, traveling back in time, by November 2009, 23 % of US consumer mortgages were underwater, meaning that they were worth less than their loans.

7:46And I think we're eventually going to see that specifically with venture capital valuations, by the way. I think there's going to be a point when it's like 90 % of AI startup valuations are going to be way lower or nil, actually. And I truly think the subprime AI crisis will be much, much worse for the Valley. As AI companies made up more than 50 % of venture capital investments in 2025, I actually don't know how any of them make it. This is something that I think about a lot. My thesis basically says these companies are all dying. I don't see how it works out. They're not getting acquired. They can't go public because they have the worst economics of all time.

8:26And the moment you show the markets that it looks real bad minimax ai company in china that went public i think it's 50 something billion dollars of revenue and like 200 something million in losses a little bit like making real money nevertheless i really don't know how that works i truly don't i sit and think about it i'm like is there a way they could get acquired is there something else they could do there really isn't all they could do is jack up their prices and hope that people don't leave and i i think that's what's going to happen if they even bother and if they can even get that far. Now, I know what you're all thinking, and I hear this a lot.

9:00I'm kind of tired of hearing it, if I'm honest, and I get why people do it, but despite the subprime comparison, this is not a too-big-to-fail situation. And in this week's premium newsletter, I'm going to dig into that question in depth. Please do subscribe. The money goes directly to me. It's my main source of income now, but it's very important to me. But nevertheless, I'm going to give you a sizable preview here because I want this information out there.

9:55We'll be right back. work. It screens thousands of stocks, builds a one-of-a-kind index, and lets you backtest it against the S &P 500. Then you can invest in a few clicks. Generated assets are like ETFs with infinite possibilities, completely customizable and based on your thesis, not someone else's. Go to public.com slash podcast and earn an uncapped 1 % bonus when you transfer your portfolio. That's public.com slash podcast. Paid for by Public Investing. Brokered services by Open to the Public Investing, Inc., member FINRA, and SIPC. Advisory services by Public Advisors, LLC SEC registered advisor.

10:23Generated assets is an interactive analysis tool. Output is for informational purposes only and is not an investment recommendation or advice. Complete disclosures available at public.com slash disclosures. Shake it up with Vital Proteins Collagen and Protein Shake. It's a high quality ready to drink shake with 30 grams of protein and 10 grams of collagen to support healthy hair, skin, nails, bones, and joints. With zero grams of added sugar, no artificial sweeteners, and absolutely no carrageenan, It's a clean, delicious way to fuel your day. So you don't just age gracefully, you age powerfully.

10:57Vital Proteins. Stay vital. Learn more at VitalProteins.com Apple Vacations, where your story starts. Spring is the perfect time to bring everyone together. Spring into savings with Apple Vacations. America's favorite vacation company. Book by April 9th and take up to$400 off international and Hawaii vacations. Or up to$125 off domestic getaways. Plus, kids stay free at select resorts when you travel through December 15th. From Cancun to Jamaica to Punta Cana and beyond. Reconnect in paradise. Plan your getaway at applevacations.com or contact your trusted travel advisor. Apple Vacations, where your story starts.

11:33And now, The Laundry Hero, brought to you by Grand Appliance. Oh, really? Okay, I'm on it. Uh-oh, what's up? Yeah, laundry's piled up and my washer just died. That's a bummer. Check out grandappliance.com. My brother got a great deal on a Maytag washer. Grand Appliance? My parents shopped there. Mine too. Whole family. I'm on the site. Wow,$5.97 with next day delivery? Problem solved. Shop Grand Appliance. Appliance experts since 1930.

12:07So, Too Big to Fail refers to the troubled asset relief program, TARP, over$400 billion, and specifically the bailouts of AIG and the surrounding finance industry, mostly focused on the commercial paper loan industry that kept major banks and financial institutions going as well as fucking GE Capital. Oh, and also the PDCF and TSLF finance systems that kept them going too. I'll get to that in a bit, but the bailout was in the trillions. Like, I don't think people realize how and why it happened. So I'm going to tell you. Anyway, AIG was too big to fail because it had$20 billion in outstanding commercial paper and several life insurance subsidiaries that were billions of dollars in the hold, thanks to AIG literally gambling with the money, not actually at casinos, on CDS's CDOs and the like.

12:52The knock-on effects of AIG's collapse would have been catastrophic for the money market funds that held commercial paper from multiple companies. And commercial paper is a short-term loan type thing, anywhere from one to four days to like, I think 200 and something days. But most of it was either very short-term and in AIG's case was also mostly without any collateral in any way. In any case, that used to be the primary way that banks and finance institutions, GE Capital, actually used to be able to fund their businesses. They would go to the market and go, hey, why don't you lend us some money just for like a few days?

13:27We'll get it right back to you. And for the most part, it worked right up until it didn't. I also want to be clear about what the bailout actually bailed out, because I think people here are too big to fail and they're like, right, they bought the houses that were being foreclosed. No, no, no, that's not what they did. It was like 40-something billion that was meant to go to stop foreclosures. No one really knows what happened to it, which is cool. Now, the vast majority of the bailouts were for financial instruments, not houses, not apartments, not helping regular people, but making sure that the toxic financial products associated with the finance industry were absorbed and bought off and that liquidity remained in the market via the primary dealer credit and term securities lending facilities that provided as much as$100 billion to banks and financial institutions a day.

14:14Now, these still exist, the repo facilities still exist, but are not used at the same scale. But this, I want you to think about this. Like every day,$50,$100 billion was just used to give them short-term funding to just get through the day. People should have been in fucking prison, to be clear. People should have gone to jail for this. But this was necessary to stop the financial system crashing, to actually falling apart. Lending, insurance would not have got funded in the same way. There was basically a credit freeze anyway, but I'm talking nothing would have happened. And even during the great financial crisis, mortgages still were getting written, loans were still getting written, things still happened.

14:55And the financial system is a confidence game. And the PDCF and TSLF existed to stop everything from actually going to zero. And like I said, those few loans were still being written and funded and they needed to make sure that insurance and borrowing and they still actually happened because the world runs on debt and insurance and oh my god this whole really reading about the finance the great financial crisis really did black pill me all over again though because you read what these fuck nuts did they were just they were like yeah we're gonna hedge our bets on the on the mortgages by buying uh credit default swaps that we assume that AIG will be able to pay, but you don't need to check.

15:35Well, they should have checked, so they couldn't fucking pay. Anyway, some estimate the true size of the bailout when you account for that liquidity to be over$10 trillion, because the under-discussed part of the AIG bailout was that the commercial paper lending industry that funded most of the finance industry kind of broke. It took years. The finance industry actually had to unwind their dependence Sonic Colgate used to use commercial paper. It's fucking weird. In any case, there really aren't any comparables to the AI bubble. While OpenAI and Anthropic might be very prominent and very annoying, their actual economic existence is relatively small, and the overall AI industry barely had$65 billion in revenue last year, with much of that flowing between a few counterparties and funded by venture capital dollars.

16:19And also, we don't know if all of that was cash, by the way. A chunk of that is the inference spend with Microsoft. Don't know if That's still tokens. If you work at Microsoft and you want to talk to me about the finances or any of these companies, please, please hit me up. I would love to hear from you. Easy at betteroffline.com. Nevertheless, had AIG defaulted, it would have left multiple banks without the funds to continue functioning and killed its insurance companies in the process, sending them into government receivership that wouldn't have guaranteed the policies had the same value or even been fulfilled.

16:53Even with that bailout, the government had to plug a hole in the side of the finance industry for several years, and they ended up bringing it back during COVID as well. While the collapse of OpenAI and Anthropic might tank parts of the market, too big to fail is a term that refers to something that has systemic risk to the economy. And these are two very different things, because the market still shat its pants during the great financial crisis. That happened. The government wasn't stopping the markets from collapsing by funding the markets. They were keeping the banking system alive so the markets just functioned at all.

17:26And I really need you to know that difference because it's not the same thing. The collapse of AIG would have quickly ripped through the funding mechanisms of most of the US finance industry. And its collapse still materially harmed one of the main funding mechanisms of the US economy. There is no such comparison with AI. AI is not producing productivity benefits at scale. It is not replacing jobs at scale. Not that I want that to happen, of course. it's not load-bearing in an economic sense in that the flow of money in the u.s economy would collapse if open ai or anthropic did and a bailout in general exists to solve a problem rather than pass the bug and i also want to be clear if that if they were truly too big to fail i don't know i know wouldn't it wouldn't it be that when anthropic goes down or chat gpt goes down the economy would stop because that's what would have happened if AIG died god damn have I nearly said AGI like five times anyway I also want to address that Fannie Mae and Freddie Mac massive mortgage companies that were absorbed by the US government they are not a comparison as the deaths of those companies would have actually destroyed the US housing market like they were like 65 billion dollars in the whole it was very very very bad like they I need you like the big short is a great movie.

18:41It does not do justice to how fucked things would have been. I also want to be clear that none of these companies should have been allowed to function in the same way. Any of the banks involved in this, none of the executives in question should be allowed to work in finance anymore. I think that it was genuinely evil. And the whole situation was caused by an industry-wide just gambling industry. That's what it was. It was truly horrifying. And everyone involved should be in jail or jobless. I think they're fucking awful. Nevertheless, it was necessary to do because of the fuckwit nature of the whole thing, because of the stupid gambling they did, because of the low reserves they had and how brittle everything was, which is their fault.

19:23That's very different to the bailout not being necessary though. Now, getting back to AI, without OpenAI and Anthropic, the AI industry would disappear, as would the demand for AI compute. It would lead to billions of dollars of loans going unpaid and the value of venture capital would probably be cut in half or as much as 80 or 90 percent, as happened with the dot-com bubble. It'd be horrible for the markets. Core Weaver would die, Iron would die, and Hebbius would die. I think the OpenAI and Anthropics death would end up ripping through the entirety of Silicon Valley. I think that it would be a payback for years of focusing on growth rather than creating anything.

19:58On the banking side, I do think we are somewhat safer. I think while there aren't heavy reserve requirements, there are actual regulations around high-risk speculation, and the scale of speculation here is teeny tiny compared to, I think it's like five trillion in synthetic CDOs or something of that nature. There aren't trillions of dollars of speculation here. Billions? Hundreds of billions? Sure. And I'm fairly sure, fairly confident, that a lot of these banks and private equity firms and private credit firms have some degree of reserves built in for these loans going tits up in the case of AI data centers.

20:33their deaths these ai companies deaths wouldn't create a calamity that would stop loans in general or insurance in general from being collateralized or any other basic functions of the u.s economy i'll also add the nvidia and the rest of the magnificent seven aren't going to die as a result of the ai bubble bursting they're not going to need bailouts i think oracle's bailout chances are there but even then i'm kind of hesitant to say so because i don't know what they'd be better that may be bailing out the loans that create a little Larry Ellison-shaped tarp just for that fuck-not. But I actually don't know if that would happen.

21:06Tarp was deeply unpopular. It was deeply unpopular. Deeply, deeply unpopular. And Trump's already having enough trouble. I'm sure he's doing the we're-touching-the-stove competition. But at the same time, I just don't see it happening in the same way. And even then, I still think Oracle would die and get absorbed into another company, probably Microsoft. Wouldn't that be funny? Man, the antitrust would be crazy. that being said i think that the ai bubble will eventually be seen as a smaller precursor much like the dot-com bubble was the great financial crisis to a larger financial crisis created by private credit and private equity in their hundreds of billions of dodgy investments in software and random companies i am still working out how bad this would be and i don't want to be an alarmist i actually don't think it will be as bad as the great financial crisis but that is a question i'm going to try and answer in the future and open ai and anthropic aren't getting bailed out it's not happening i don't care if you think oh the department of defense uses them they they're not going to bail them out for that reason they're not going to put them into conservatorship it's not going to happen it's not it would be deeply unpopular and also to what end a festering hole in the side of the government we already have congress anyway ai data centers aren't getting bailed out either i really need to be clear about that houses were not bailed out foreclosures were not bailed out.

22:22Perhaps the ABSs might have some hope, the asset securities, but even then I don't think the scale of those is anywhere close to the great financial crisis. And to be clear, the collapse of the AI industry will fucking hurt. It will brutalize the banks and private credit firms involved. It will fuck venture capital for years, if not over a decade. Earnings seasons are going to look like the dog from John Carpenter's The Thing. It's going to be horrible for them and horrible for people invested in the market. It's not going to be nice. I am not saying that it's not going to hurt. But that is meaningfully different to anything being too big to fail.

22:59And I'm tired of people saying things without actually bothering to understand what they mean.

23:40And now, the No Panic Party Save, brought to you by Grand Appliance. Your party is safe. Wow, Amy, great party. And the food? Incredible. Thanks. And did I tell you my stove died two days ago? What? Did you panic? Nope. I called Grand Appliance and got a great deal with next day install on the Frigidaire gallery I wanted. Next day? Wow. GrandAppliance.com, right? That's it. My family's shopped there for decades. Shop Grand Appliance. Appliance experts since 1930. Hello, it is Ryan, and I was on a flight the other day playing one of my favorite social spin slot games on Chumbacasino.com. I looked over at the person sitting next to me, and you know what they were doing?

24:20They were also playing Chumbacasino. Everybody's loving having fun with it. Chumbacasino is home to hundreds of casino-style games that you can play for free anytime, anywhere. So sign up now at Chumbacasino.com to claim your free welcome bonus. That's Chumbacasino.com and live the Chumba life. Sponsored by Chumbacasino. No purchase necessary. VGW group void. We're prohibited by law. On 21 Plus, terms and conditions apply. Today, we're exploring deep in the North American wilderness among nature's wildest plants, animals, and cows? Uh, you're actually on an Organic Valley dairy farm where nutritious, delicious organic food gets its start.

24:57But there's so much nature. Exactly. Organic Valley small family farms protect the land and the plants and animals that call it home. Extraordinary. Sure is. Organic Valley, protecting where your food comes from. Learn more about their delicious dairy at ov.coop. This is an iHeart Podcast. Guaranteed human.

From the publisher

In this week's Better Offline monologue, Ed Zitron breaks down why AI isn’t too big to fail, why no bailout is coming, and how people need to learn more about history before using it to justify burning billions of dollars.

This week’s free newsletter: https://www.wheresyoured.at/the-subprime-ai-crisis-is-here/ 
Premium newsletter tie-in out later today - save $10 off a year of my premium newsletter: https://edzitronswheresyouredatghostio.outpost.pub/public/promo-subscription/gzqwkv54e1  

YOU CAN NOW BUY BETTER OFFLINE MERCH! Go to https://cottonbureau.com/people/better-offline and use code FREE99 for free shipping on orders of $99 or more. Buy our new “FUCK DATA CENTERS” shirts today!

---

LINKS: https://www.tinyurl.com/betterofflinelinks

Newsletter: https://www.wheresyoured.at/

Reddit: https://www.reddit.com/r/BetterOffline/ 

Discord: chat.wheresyoured.at

Ed's Socials:

https://twitter.com/edzitron

https://www.instagram.com/edzitron

https://bsky.app/profile/edzitron.com

https://www.threads.net/@edzitron

Email Me: ez@betteroffline.com

See omnystudio.com/listener for privacy information.

More from Better Offline

All 276 episodes
Monologue: AI Isn't Too Big to FailBetter Offline · 18 min
Listen in VO