- Big Business This Week
- Posts
- Making Sense of Anthropic's Blockbuster IPO
Making Sense of Anthropic's Blockbuster IPO
Plus: GM CEO Mary Barra Exits Under Pressure
BROUGHT TO YOU THIS WEEK BY:

Huge doesn’t begin to describe it. Anthropic $ANT ( ▲ 17.07% ) went public on Thursday, selling 980 million shares at $83 a piece, raising $81.34 billion and valuing the company at $1.4 trillion. That’s a little below the $1.4 to $1.5 trillion that many analysts had projected, but it’s still the largest IPO in history. Shares jumped past $111 million on Thursday’s trading debut to raise the valuation near $2 trillion, eclipsing SpaceX’s $SPCX ( ▼ 3.53% ) $1.75 trillion IPO-day valuation. That’s a big shot in the arm for a market that’s been spooked by the Nvidia $NVDA ( ▼ 3.3% ) shenanigans, worries about AI’s circular financing, and the growing public distrust of AI and data centers.
There have been some loud groans from investors who missed their Anthropic payout, including collapsed FTX crypto exchange boss convicted fraudster Sam Bankman-Fried. FTX took a 15.5% stake in Anthropic for $580 million in 2022. It would have been worth $37 billion at the IPO, but was liquidated for a mere, trifling $1.34 billion in 2024. Anthropic’s seven founders will each end up with about $18 billion, most of it still tied up in the company. Another winner from all of this? Open AI $OPEAZZX ( ▼ 0.74% ) , the ChatGPT owner, which has recently been putting off plans for its own IPO, amid investor fears that OpenAI may have missed its window. Now CEO Sam Altman says an IPO is really really gonna happen, “at some point,” without offering specifics. “These recent months have been a newly challenging period,” Altman said at a business conference in New York. To better understand the value of the Anthropic IPO, and how it positions the AI ecosystem, Big Business This week columnist Peter Green spoke with Lior Prosor, a managing partner at Deep33, a venture fund investing in compute, advanced energy and autonomous manufacturing.
Peter Green: Let's start with the big picture. What does the Anthropic IPO mean at this moment?
Lior Prosor: The few things to think through in the Anthropic IPO are as follows. Number one, which is pretty unbelievable: this is a business that's reached 80-something billion dollars worth of run-rate revenue, and they've spent less than 10 billion dollars to get to that run rate. They still have about a hundred billion dollars in their bank that they've raised since the foundation.
The company is going into this IPO with a very strong balance sheet, and they are approaching it in a very smart way, which is: We almost don't care about the initial price of the IPO. We want to make sure that we are incredibly well-capitalized in order to take advantage of the opportunity ahead of us, which they define as almost endless. It's intelligence. I think they put it at 40 trillion dollars or something. So we have to make sure that even if [compute] becomes more expensive, even if underwriting on these data centers becomes more difficult, we have the balance sheet to do it.
Now that it's priced at $83, obviously the demand was extremely high. This is a big IPO. That means a lot of capital needs to be found in order to generate that level of demand. It was highly oversubscribed and post this IPO, the business has about 200 billion dollars of equity on the balance sheet. It can support all these compute commitments, all these data center commitments that they need in order to grow the business.
It almost doesn't matter how they trade short term. I think they'll trade well. The business is now probably the most well-capitalized business in the AI infrastructure space, and they can be very, very, very aggressive in getting the amount of compute that they need in order to serve their customers and to service themselves.
One of the big questions about AI has been how durable corporate take-up really is. How much of it is companies dipping a toe in the water, rather than being willing to pay what it actually costs?
There's three core forces when it comes to enterprise take-up that you need to think through. Number one, most enterprises are relatively rational economic buyers, meaning they do not spend money unless they think that there's an ROI on the other side. Sometimes they might be wrong, but they're making that decision with a view that there is ROI.
Point number two: people are using more and more tokens. That's why the growth rate of OpenAI [and Anthropic] in this market has been the way that it's been. And the price per token continues to go down 80, 90% a year for the last three years now. These two forces together [mean] there is enough ROI where these businesses think that it's worth doing it, and the price per token continues going down, which opens up more and more buyers and more and more use cases.
The third layer, and this is where [it's] very early in the game, is: Do I feel comfortable deploying AI in my environment? There's a lot of people that are not buying at scale or doing things in the core of their organization because of the risks that come with deploying these models. As security and safety measures ramp up and we get better harnesses for this technology, that adoption will continue to increase.
As tokens continue to get cheaper, as demand increases, I think it's a good sign that there is ROI for customers and it's worth their spend. If not, they wouldn't be doing it.
How does Anthropic compare to the competition — OpenAI in particular, but also Google, Grok, or the Chinese models? Is there a threat to its market down the road?
The simple answer is yes. There's a very competitive market, and how much further ahead you are than others in the game is at most three, four months today. If you are not continuing to push the frontier in three to four months, you are “commoditized.” So it makes it a hyper-competitive game.
At this moment, the frontier is 100% dominated by Anthropic and OpenAI. How do we know that? Because roughly 30% or so of tokens are OpenAI and Anthropic tokens, and 80% of the economic value is coming from those tokens. People are willing to pay for the frontier, and the frontier is OpenAI and Anthropic.
What happens down the line? There's two cases to be made. Case number one is recursive learning: a model is intelligent enough to build the next version of the model on its own. That is already happening. It's also what the Doomers would say makes AI scary — machines are building the next level of machines without tons of human involvement. A big chunk of the next model that OpenAI and Anthropic are going to release is done by recursive self-learning. And there's one claim that says that if you are already at the frontier and you have the most capital, which is what OpenAI and Anthropic now have, actually the frontier extends, meaning you're able to release new and better models faster than anyone else can.
The other side is that as use cases expand, the distance between one and the other is not going to be sufficient for someone to pay a premium. And that means that the price that OpenAI or Anthropic is able to charge goes down over time. I'm more a believer in one than in two, but there's a lot of people that believe in two as well.
It's hard to know whether these things become like automobiles — a commodity — or something else.
You can think of a commoditized industry where scale and distribution is meaningful enough where it's hard for others to compete with it. Take the airline industry. It's not surprising that you have three plus or minus airlines that dominate air travel. You can start another airline, but they'll end up going bankrupt at some point in time. Or take the cloud market. The reason that AWS $AMZN ( ▼ 3.83% ) , Microsoft $MSFT ( ▼ 1.42% ) , and Google $GOOGL ( ▼ 3.5% ) are so dominant is this is a game where capital is king. There's no way for you to deliver a unit of compute at the price that they can without having massive scale. So just the scale and the recursive self-learning of these models actually allows them to compound, and it will be a game of two to three major players.
What does the success of this pricing suggest for OpenAI, which hasn't gone public yet, and for the wider AI trade?
Number one, this is very positive for the rest of the AI industry, all the way from the data center builders, people that are building turbines that go into producing the energy for these data centers, different chip providers from NVIDIA and AMD $AMD ( ▼ 8.45% ) , all the way to TSMC. Number two, I think it bodes quite well for the OpenAI IPO. OpenAI didn't want to be first, and recently their growth rate has been as high, if not even higher than Anthropic. So yes, they're a step behind, but there's a lot of really good reasons why you should be excited about an OpenAI IPO as well.
[Third], at least short-term, it shows that for premium assets the market is still very interested in the AI trade, which was somewhat shaky over the last month. Certainly with Jensen leaving NVIDIA, we've moved from saying there's not a single piece of bad news about AI to pretty much most things that you'll read is about how bearish we are. There's sort of like a pendulum set to swing, and this is probably a beginning of a swing to the other side.
NVIDIA showed what key-man risk looks like. Is there that kind of risk at Anthropic — one person, one event that could knock them off?
Anthropic is in an incredibly dominant position, both in terms of their capabilities, the level of their models, their distribution, and now also their ability to continue growing their compute. Compute is not just important in order to provide a product to the end customer, it's also super important for their own internal needs of building and training their next generation models. So the Anthropic bet, putting aside the actual price, has become a lot less risky than it was 48 hours ago.
Having said that, it is an unbelievably competitive industry. If Anthropic, for some reason, is having [an] internal security [failure], [or is] unable to release their next generation model there's ways to unseat them from some of their customers, some of their use cases. But again, incredibly talented team, unbelievable product, and now a very dominant capital position going into the next few years.
And Dario Amodei is a different kind of figure from Elon Musk or Jensen Huang.
He's a very different figure. A lot of the reasons that people dislike the industry are some of the things that come out of his mouth, versus Jensen [who] was an unbelievable ambassador and evangelist for the industry, and Elon doing things that are science fiction and at the core of US exceptionalism. Dario is not any of those. Your general retail investor would not be as shaken by Dario leaving versus Jensen or Elon.
But this is the highest growing, fastest growing, most profitable company in the history of the world. I don't think that happens without a CEO that has navigated the business well. He's built up the best business [that] has ever been built in the history of capital. It's a business that went from zero to nearly a hundred billion dollars of revenue in three years or so.
It priced higher than many expected. Is there money left on the table — should the man in the street get in?
This isn’t investment advice, but If I had to guess, shorter term, based on the amount of demand, this will open significantly higher than the IPO price. But if I'm the man on the street and I have an opportunity to buy the stock and I don't know what it opens at — 100 or whatever it opens at — [with] these things [I'm] pretty comfortable waiting for another buying opportunity. We can settle down a little bit.
One word you can't avoid in the AI build-out is circularity. Is there a risk to Anthropic's valuation if the circle grinds to a halt?
A lot of this AI build-up is so expensive that the only way that it could really come to be is if some of these major participants across the stack that are very, very, very strong balance sheet players, like the hyperscalers or NVIDIA $NVDA ( ▼ 3.3% ) or AMD $AMD ( ▼ 8.45% ) , [support it]. To say very plainly, yes, a lot of these market participants are assisting other parts of the value chain to get to where they need to get to. And this is well beyond NVIDIA and Google and Amazon participating in the IPO or providing vendor financing. This is happening across the stack. If you need to build optical interconnects or small chips that are going into some of the NVIDIA systems, you're getting support from NVIDIA. They will prepay for a lot of your products. They will help you with the tooling. They are almost like the large hyperscalers — they are the organizers of this industry in order to scale the way that it has.
The question is sort of where is the circle today? Who's not making money in the current process? You have the producer of the chips, which is TSMC. They're making a lot of money. You have all the suppliers of TSMC. They're making money. You have NVIDIA that's making quite a bit of money. And then you have the model providers, which are Anthropic and OpenAI, that are making a lot of money. Everyone is making money. The only question is the end buyer who is buying the token now at X getting enough ROI on their spend. That's the only part of the equation that we have to ask ourselves, will it break at some point?
There's significantly more demand than there is supply right now and cost per token is going down simultaneously. So at some point, we will reach equilibrium. But right now, we're on the other side of that. Demand is massively outpacing our ability to supply tokens.
So barring something entirely exogenous, prices just keep going up?
Over the next 12 months, we're not going to see any let-up in GPU prices. [But] a lot of innovation happens in every layer of the stack to provide much cheaper tokens. As you move from these massive data centers that you need for pre-training into smaller data centers that you can use for inference, build-out doesn't only have to be in the US. So I actually think that supply is going to increase, the cost per token is going to continue to be driven down, and medium to long term, prices should stabilize and go down.
Last thing: some investors are concerned about the multiple share classes, the slice of each raise going to the public benefit foundation, and unequal voting rights. Does that affect your view as a shareholder?
It's a weird one. I can't say that I fully managed to understand how that [structure] work[s]. But we've seen structures that are more difficult to underwrite than this one, like founders having full control voting rights, owning the compensation committee — I'm talking about Mark Zuckerberg and Elon Musk. At the end of the day, as long as the company's performing the way that it is, [with] the dominance of the frontier, then people are not going to pay too much attention to it. It'll probably be a much more meaningful story if the company's underperforming.
(This interview has been edited and condensed for clarity and brevity)
—Peter S. Green
Big Businesses mentioned this week:
$ANT ( ▲ 17.07% ) , $SPCX ( ▼ 3.53% ) , $NVDA ( ▼ 3.3% ) , $OPEAZZX ( ▼ 0.74% ) , $AMZN ( ▼ 3.83% ) , $GOOGL ( ▼ 3.5% ) , $AMD ( ▼ 8.45% ) , $GM ( ▼ 0.72% ) , $DB ( ▼ 0.4% ) , $DIS ( ▼ 1.39% ) , $SKYD ( ▼ 4.53% ) , $NFLX ( ▼ 2.28% ) , $JPM ( ▼ 1.2% ) , $AAL ( ▼ 2.15% ) , $DAL ( ▼ 2.0% ) , $UAL ( ▼ 1.95% ) , $NTDOY ( ▲ 2.62% ) , $UNP ( ▼ 3.43% ) , $NSC ( ▼ 2.07% ) , $HIMS ( ▼ 5.04% ) , $ENHA ( ▲ 2.09% ) , $TSLA ( ▼ 2.1% )
This week, big business!
The Usual Suspects
GM’s Barra heads for the exit: After 12 years leading what was once the world’s largest carmaker (It’s now No. 4 by sales), Mary Barra says she is leaving. The news was no surprise, Barra turns 65 on Christmas Eve, the standard retirement age for GM execs, and she’s served longer than any General Motors $GM ( ▼ 0.72% ) chief in recent years. In a world where U.S. carmakers are under heavy attack from China, Japan, South Korea, she’s done a remarkable job shaking up GM’s hidebound bureaucracy, adding a bit of Silicon V alley spice by moving into autonomous vehicles, and cutting unprofitable car designs and foreign markets to refocus the company around its best-selling, high margin pickups. Shares are up more than 135% since she took over, and are up about 20% in the past year. Barra had been touted as a replacement for Elon Musk if he leaves Tesla $TSLA ( ▼ 2.1% ) , but no word yet on what she plans to do when she gets her gold watch.
Deutsche Bank’s Sewing machine exposed: A leak of 15 million internal documents from Germany’s Deutsche Bank $DB ( ▼ 0.4% ) show CEO Christian Sewing, who supposedly rescued the ailing mega-lender, built much of the bank’s success on a series of allegedly deceptive accounting tricks, sanctions-flaunting, money laundering and other failures, despite promises to reform the bank and distance itself from unsavory oligarch clients in the U.S., Russia and Latin America. Among the revelations in a multi-part investigation published in the Financial Times: Deutsche processed $600 million of suspicious payments linked to Jeffrey Epstein; Sewing pushed to expand business with Russian oligarchs; Weakened financial controls let Deutsche allegedly launder money for drug cartels; The Bank accidentally leaked Donald Trump’s year-end tax statement and declined to let him know. The Frankfurt-based bank told the FT: “We are a different bank today. Under Christian Sewing’s leadership, Deutsche Bank has gone through a fundamental transformation.” That was news to markets, as Deutsche shares fell 6% when the revelations were first published.
Mouse Trap: Disney’s $DIS ( ▼ 1.39% ) the latest entertainment giant to promise massive annual savings by cutting jobs and investing in AI, while boosting spending on entertainment. The company says it will hit $3 billion in annual savings by fiscal 2029 in an effort to better compete in the global streaming market. Content spending will rise 9% this year to $26 billion, tripling its production budget outside the U.S. Last week, Skydance/Paramount/Warner combo $SKYD ( ▼ 4.53% ) said it planned to reduce annual costs across its brands by $6 billion, and last month Netflix $NFLX ( ▼ 2.28% ) cut 1,000 jobs. Disney shares are up 15% in the past year.
Who pulled the plug? What’s going on with the internets? A series of financially disastrous tech outages have hit the country in the past several weeks, including the big air traffic control outage that grounded all the flights in the U.S. for several long hours. Among the victims: hospitals, Microsoft $MSFT ( ▼ 1.42% ) , Chase Bank $JPM ( ▼ 1.2% ) and the top-secret National Security Agency that runs America’s electronic eavesdropping. Microsoft says a server fix caused the glitch that cost $7 billion in lost workplace productivity, insured cuber losses of $700 million and a potential demand for poor-service credits of as much as $800 million a day for the 3-day outage. The air traffic control outage that the FAA blamed on routers in Texas and Arkansas cost airlines about $8.25 billion, according to Raymond James, or about $354 million for American $AAL ( ▼ 2.15% ) ; $244 million for Delta $DAL ( ▼ 2.0% ) , and $229 million for United $UAL ( ▼ 1.95% ) . And that’s not counting the knock-on effect to businesses that rely on the airlines and to the passengers themselves, for expensive airport meals, taxis, rental cars, and missed work. JPMorgan blamed an internal software bug, but didn't put a cost on the hours-long processing delays. President Donald Trump told FOX News that people should “relax” about the outages on Wednesday.
ADVERTISEMENT

Most firms sell a product. We build plans and take positions. Keebeck is a modern family office for founders and families who want clarity, speed, and alignment.
We co-invest our own capital alongside clients, so the incentives stay aligned. One transparent fee. No product stacking. AI powers our research, cash-flow and tax modeling, and risk monitoring, so you get answers in real time and decisions that hold up in the real world.
Keebeck translates complex data into simple direction and stands next to you through the full life cycle of an investment, not just the pitch. Public markets. Private deals. Estate, tax, and governance. Liquidity when it matters.
We measure success in outcomes, not AUM. We show our work: every model, every assumption, every fee.
If you want spreadsheets, hire a software developer. If you want a partner with skin in the game, call Keebeck Wealth Management.
Keebeck Wealth Management. Helping you become the CEO of your capital.
END OF ADVERTISEMENT
The Short Stack
Hey Kid - Wannabuy some Pokemon cards? It’s not quite Picasso, but Pikachu, the classic Pokemon card is turning into a real collectible. 30 years after Nintendo $NTDOY ( ▲ 2.62% ) first introduced the game and the playing cards, Pokemon traders are making a living finding rare cards in old kids’ collections and selling them for millions. Pro wrestler Logan Paul sold a 1998 Pikachu card for $16.5 million at a February auction. “It is very similar to trading stocks,” collector Scott Pratte told the Wall Street Journal. “I think Pikachu and the cute little Pokémon on the cards distract from how competitive and cutthroat it can be.” It’s not clear what the percentage rise in Paul’s card’s value was, but it certainly must have been the 465% rise in Nintendo stock since it was listed in New York in 1998.
Loose lips sink trains? Union Pacific $UNP ( ▼ 3.43% ) railroad chief Jim Vena may have sunk his company’s planned $71.5 billion takeover of rival Norfolk Southern $NSC ( ▼ 2.07% ) railroad, after video surfaced of Vena threatening to raise shipping rates of anyone who opposed the merger. The U.S. Surface Transportation Board, which has to approve the merger, said it had “serious concerns” that Vena’s threats had silenced opponents of the deal, including farmers, chemical companies and other freight shippers who fear the merged railroad will raise rates. Union Pacific said Vena is a “dynamic speaker,” and meant nothing by it. A combined rail network would stretch the breadth of the U.S.
Make America Bulked Again: Telehealth and beauty groups are rushing to cash in on the peptide craze after Health and Human Services Secretary Robert F. Kennedy Jr. said he won’t take enforcement action against compounding pharmacies that sell six unproven peptide injectables, opening what could be a $35 billion market in the next five years. News they won't need regulatory approval for sex stimulants and muscle growth drugs has prompted the promotion this week of new supplements from online supplement companies including closely held Noom and Hims & Hers health Inc. $HIMS ( ▼ 5.04% ) . Enhanced $ENHA ( ▲ 2.09% ) , backed by Donald Trump Jr., began offering the six experimental peptides within a week of the announcement. Shares in Enhanced are down 91% this year.
About that ever-impending merger: Speculation that Tesla $TSLA ( ▼ 2.1% ) and SpaceX $SPCX ( ▼ 3.53% ) might merge is rife again, after Elon Musk was asked about a merger at a New York Times business conference. "Would they be better together? I mean, probably yes," Musk said in an interview in New York. The New York Times published a story detailing how the two companies work hand in hand, and citing employees who think the two should end the separateness charade and formally unite. But Elon’s not straying from his oft-repeated comments that it’s ultimately up to shareholders and could — and maybe should — go either way. “Some people out there assume that a SpaceX and Tesla merger is a certainty,” Musk said last month on a call with investors. “I think that would be overstating it.” Any deal would have to win regulatory approval which will be much harder come January, when the Democrats take control of Congress. But Musk already treats the companies as though they were subsidiaries of one firm, and the line between them is blurred. “Even in the days of John D. Rockefeller, you didn’t have this kind of consolidation of wealth and power.” said Jeffrey Sonnenfeld, the senior associate dean for leadership studies at Yale School of Management.told the New York Times. Tesla shares fell 1% and SpaceX shares were down a bit more than 2% in midday trading on Thursday.
Want more Cheddar?
You’re clearly into smart people talking about even smarter things. Lucky for you, that's literally our whole deal at Cheddar. We interview the brightest minds in business, finance, and tech. If you'd like more in-depth analysis from interesting people, lcheck out our where to watch page and turn us on 24/7! Your wallet will thank you and so, more importantly, will your mind. But also your wallet. Remember that.
Peter S. Green is a veteran reporter and editor who has spent more than two decades covering business and finance from Eastern Europe to New York City, and has worked for Bloomberg News, The New York Post, The New York Times and The Messenger. He lives in New York City and is always looking for the next big story. Email him here.