Will Anthropic's $2B IPO Fly Or Stink?

Plus: Goldman’s new gold man as new CEO is mooted

Author

Cheddar's Big Business This Week
October 01, 2026

In partnership with

Anthropic’s $ANTHZZX ( ▲ 3.42% ) planned IPO is making news, after a copy of its S-1 filing (a preliminary document filed with the S.E.C. ahead of a public offering) was leaked to several news organizations. It appears to be an extraordinary document that includes 80 pages of warnings about AI’s potential existential risks, including about how AI could destroy the world and notes that AI poses “existential risks to humanity.” Yeah okay, mass extinction events have their place in the world, but if we all miraculously do survive past Christmas, what will the IPO do to the markets, and is it smart to be buying a piece of Anthropic now? Big Business This Week’s Peter Green spoke with Tal Elyashiv, a venture capital veteran who’s launched or invested in several AI companies, and has taken a close look at the available data from Anthropic. They spoke about the numbers, the risks and what the deal means for the rest of the AI market.

Anthropic's IPO would value the company at $2 trillion. Why will it fly, or why will it sink?

First let's look at the numbers. $2 trillion is a major, major signal. It's the biggest IPO ever. It's the first business-to-business AI company to go public. Their revenue projection now exceeds $100 billion annualized for this year, so you're really looking at 20x in terms of valuation. Both numbers are high, but not astronomical. If you look at Google $GOOGL ( ▼ 1.7% )  at its peak or Microsoft $MSFT ( ▼ 0.02% )  at its peak, there were  12x, 13x, but none of them showed growth of 1,000% like Anthropic showed.

What about spending?

Anthropic burned through $8 billion last year, with revenue of $4.6 billion. Over the next decade, they’re planning to burn through $500 billion-plus. That’s a big, big, big jump in spend, and that's a number that investors need to understand. It reflects the cost of training and operating frontier models at scale. But it also means that the path to sustainable margins is long and very cap-ex dependent. The second quarter of 2026 showed for the first time positive adjusted operating income. It's a reversal of the previous trend of losses. But it's a single data point and it's a single quarter. I think everybody will watch Q3 and Q4 numbers to see if they're seeing a trend.

Amazon $AMZN ( ▼ 0.37% )  and Alphabet $GOOG ( ▼ 1.63% )  are the two largest shareholders. How does that affect Anthropic’s future?

Amazon holds about 21% of Anthropic, and Alphabet owns about 15%, and both of them are infrastructure providers for Anthropic, so Anthropic spends a lot of their expenses on those companies. What percentage of the spend is actually going to those players, and how it's being priced [is] really significant to what story you tell about growth. There is no comparison for Anthropic. When Google went public, they were not making anything like that kind of money (Google’s S-1 reports $105 million in net income for 2003, before its 2004 IPO). Here they're IPO-ing on real revenues and 1,000% growth from previous year's revenues. That's unparalleled. But it's a short history. You can't scientifically extrapolate from this. So you tell a story about the future as an investor. Nobody can tell what's going to happen two years from now.

With one quarter of real profit, it's a coin toss.

Exactly. There's no way to extrapolate with one point of data.

So is $2 trillion rational?

It's high, but it's not totally irrational. What it is, is, it's extremely demanding. Can they really win and hold enterprise clients and government clients? Because of [Anthropic’s] security capabilities, this is where a lot of the premium should come from because. If you look at the S-1, about 80 pages there talk about the “existential threat to humanity.” Their platform provides safeguard layers and the Mythos tier. They're all products that are built for large enterprises and government. The question is can they keep those clients and grow that market?

Can they? What are the biggest risks?

One is the gross revenue versus net revenue and the accounting around that. The other one is there is a litigation going on with the Department of War [Anthropic is suing the U.S. Dept. of Defense, which barred it from government contracts after Anthropic said it would not build fully autonomous weapons or mass domestic surveillance]. This may have a significant impact down the road. People think about competition risk as Anthropic versus Gemini or OpenAI $OPEAZZX ( ▲ 1.27% ). That's not the real competition risk. The real risk is: Will a bunch of their clients over time realize that they can work with cheaper models and still get the same value without paying the premium?

Such as industry-specific models kept on their own servers, air-gapped from the rest of the world?

Exactly. That's an unknown. The other thing is not really a risk, but it's a signal that should cause people to pause and think: Insider ownership at Anthropic stands at around 3.5%. Kind of rare for a company that young. They raised a lot of money and there was a lot of dilution. But as an investor you need to ask yourself what's your comfort level.

What's the bull case?

If you believe that Anthropic will continue growing exponentially then, at $2 trillion, which is a very aggressive number, you're paying for the possibility that maybe Anthropic will become a $5 to $10 trillion company in the next few years. Otherwise, it's going to be very hard to maintain that valuation.

That reminds me of Krispy Kreme $DNUT ( ▲ 1.22% ). It had triple-digit growth, imagining the market would keep expanding, but Americans just didn't want to eat that many donuts. Then the GLP-1s came along! Can Anthropic keep scaling?

At this valuation and this multiple, you have to show significant exponential growth over time because otherwise it doesn't hold. At 20x current revenue, with [ongoing] litigation, $500 billion in planned infrastructure spend, and one profitable quarter, you have to watch the next two quarters really closely. If both of them are showing significant growth, then in the short term, you'll see significant increase in Anthropic shares. However, if the third and fourth quarters don't support that, I think the market will be very disappointed. It will overreact downwards.

Like Krispy Kreme, that depends on demand. Will corporate customers keep paying for all this usage?

[Companies are] starting to audit whether the metrics [justify] usage. This is something that has nothing to do with the quality of product of Anthropic. But I don't see it affecting use in the next two quarters. My youngest daughter works as a data analyst, and her company allocates [a set number of] tokens a month. If you do more than that, you have to justify it. Most companies don't do that. When I'm developing AI agents and sub-agents, and I send them to do complicated tasks, maybe it's not worth it to the company. Nobody's asking.

Are the labs selling below cost?

I'm a great believer in the industry, and I think cost in the industry over time will go significantly down. They know that costs will go down over time, so they can afford to lose money now. The question is how fast does it really happen? You see significant competition from Chinese companies with much cheaper models. But the next generation of U.S. models could be less expensive than the Chinese ones. Costs are going down, there's no doubt in my mind.

Oura just said there is market uncertainy, but are markets running out of cash to invest in firms like this?

I don't think so. Cash in the market is really a function of the psychology of investors. If you are confident in where the economy is going, if you're confident in growth over the next year or two, then there's a lot of money in the market. If you're concerned, and there's a lot of reasons to be concerned right now, the war in the Middle East, the issue of oil, the Strait of Hormuz, the bubble in  China, [the tension between] Russia and Europe. There are a lot of tensions that can escalate. Nobody knows what the impact would be. And the markets are very high, so from an investor psychology perspective, people are not sitting here and saying, ‘I know the next few years are going to be amazing.’ They're asking questions and they're very selective. It's not that there isn't money; it's going to safer things. With interest rates going up, suddenly, you have the ability with municipal bonds in the U.S. to make almost 7% on zero risk. Still,  the Anthropic story is really exciting. It's attracting people psychologically because of what Anthropic is, and because all of us touch AI a little bit. So even if we can't understand what it means for humanity, the interest is there. But at $2 trillion, they have to show incredible success over the next two quarters to maintain this multiple and continue to grow. 

What does this IPO mean for the rest of AI?

It creates a benchmark. If you're a CEO of a startup, you have to explain why you're not showing this trajectory. It becomes harder to raise money for companies that are not making money in the AI space. It kind of closes the window of investing in private frontier lab companies. You have Google, Anthropic, and OpenAI. Who in their right mind will invest in a startup that is trying to compete with them? I think the new money will go to vertical AI companies [where] it's much easier to generate real revenues early in the life of a company. This is where investments are going to go over the next five years. For investors, those models are attractive because they have a significant moat in specific industries with proprietary models, proprietary data, proprietary knowledge.

(This interview has been edited and condensed for clarity and brevity).

—Peter S. Green

Big Businesses mentioned this week:

This week, big business!

The Usual Suspects

  • Where you gonna find that triple-shot oat milk latte? Starbucks $SBUX ( ▲ 1.01% ) is sacrificing some 250 poor-performing cafes on the altar of the profit-or-perish turnaround plan of $100-million, jet-commuting CEO Brian Niccol. It will take a $300 million charge, including severance payments and lease exits. Last year Niccol closed about 600 laggard coffee shops, but the company still has about 18,000 outlets in North America. The plan seems to be working. Sales at stores open more than a year were up a combined 8.1%, and shares have climbed nearly 12% in the past year.

  • Goldman’s new gold man: Goldman Sachs $GS ( ▼ 0.41% ) CEO and DJ-in-Chief David Solomon could be out in a year, and the Wall Street Journal says the new guy is likely to be John Waldron, Goldman’s president. Waldron, 57, got an $80 million retention bonus last year, when the board heard he was talking to Apollo Global Management $APO ( ▼ 1.48% ) about a job there. Waldron earned $45 million in annual compensation last year, while Solomon had a $47 million, and the two have been working on a “One Goldman“ strategy to ensure the firm’s business units keep steering business to each other. Oddly, after the elite New Jersey prep school Lawrenceville Academy, and an English degree from Middlebury College in Vermont, Waldron was turned down by Goldman and went to work for Bear Stearns, which collapsed in the 2008 mortgage crisis. He joined Goldman in 2000. Goldman’s shares have quadrupled in value under Solomon.

  • DoorDash Dressup? Wanna get fancy for your food? Now you can order dinner and your drip together. DoorDash $DASH ( ▼ 1.05% ) announced agreements with a host of clothiers and retailers and clothing brands to deliver fashion alongside food. New partners include Timberland, Vans, Anthropologie, Macy’s $M ( ▼ 0.04% ) and North Face. “Hopefully, I’m going to reach a broader audience and a different customer base,” Macy’s Max Magni, chief customer and digital officer told the Wall Street Journal. It will offer two-hour delivery from some 360 of its 400+ stores. Macy’s shares are up a little over 4% this year, as department stores struggle against online rivals and discounters. 

  • Wild blue yonder: First the bad Boeing $BA ( ▲ 3.35% ) news: A software glitch in the latest 737 MAX jet, the MAX 10, has delayed FAA safety certification of the plane, the last model of Boeing’s best-selling airframe, which began design in the 1960s. Boeing’s turnaround under CEO Kelly Ortberg, who took over two years ago has hit repeated turbulence, including strikes, and slower-than-expected recertification. Shares are down nearly 28% in the past year. On the plus side, the U.S. Navy picked Boeing for a $20 billion contract to build the next generation carrier-launched fighter jet, replacing the F/A 18 Hornet, which first flew in 1978. The new plane should be ready within a decade.

ADVERTISEMENT

The ice cream shop that makes money when it's cold

28 Wishes sells ice cream in Los Angeles. Below 70°F, sales fall about 20%. The weather is out of their hands. Rent isn't.

So the owners started putting about $20 a day into Kalshi weather markets, taking the cold side. The days that keep customers away now pay something back.

This is hedging. Big companies have done it for decades, buying protection against bad weather, fuel spikes and rising rates. It used to take a broker, a trading desk, and an order size no corner shop could meet.

Kalshi opens it up. Contracts on weather, fuel prices, inflation, tariffs and regulation, starting at a few dollars. Take a position on the outcome that would hurt you. If it hits, the payout softens it. If it doesn't, the contract expires and the good month was the point.

END OF ADVERTISEMENT

Tech Turmoil

  • First, Do no evil. Then cash out! OpenAI $OPEAZZX ( ▲ 1.27% ) CEO Sam Altman says he’s delaying an IPO for the company until the $850-billion company has solved its safety issues. “We’re going to prioritise the mission and safety and make sure that we can very confidently scale to the next stage of AI without people debating what percentage chance we’re going to do all these bad things in the world,” Altman told journalists on Tuesday. OpenAI bots have evaded human and robotic controls to hack into companies from AI firm Hugging Face to a series of government websites, said OpenAI, without the knowledge of company officials. Along with a mildly incoherent plea in September by Palantir $PLTR ( ▲ 1.67% ) CEO Alex Karp for the government to take over AI firms because of the damage they could do, that’s prompted Florida’s populist GOP attorney general James Uthmeier to ask a Florida court to ban OpenAI from developing new tools without third-party safeguards and to block minors from using its chatbot. “We’re committed to working with Florida and other states on advancing pragmatic AI policies that apply to the entire AI industry – not just one company,” an OpenAi spokesman said. The lawsuit notes that the industry itself has asked for the government to rein it in. All Florida wants, said the suit, is: “No new model development without independent safety guardrails; no more harvesting children’s data; no more calling this product safe, accurate, or reliable; no more pretending it’s human; no more tricks designed to keep users talking past the point of danger.”

Instagram Post

  • Only fools pay taxes. And the tax geniuses at Mark Zuckerberg-led Meta $META ( ▲ 0.1% ) have taken that to heart, claiming a research and development tax credit that labels AI data centers as "experimental pilot facilities." The tax credit is meant to help spur innovation by giving companies writeoffs for taking risks to innovate. Meta saved $3.9 billion in 2025, cutting its IRS bill by over 70%. That’s one company accounting for 10% of the claimed credits. But Meta’s not so sure it’s on solid ground claiming a $50 billion data hub is an experiment. It’s boosted reserves against an IRS audit by 45% to $18.74 billion. Meta shares are up about 10.5% in the past year. 

  • Circularity Watch: Nvidia’s $NVDA ( ▲ 1.09% ) discovered the secret to prosperity, even though it’s not much of a secret any more: Lend to your buyers. But with lead times on new data centers approaching two years for key components including generators and transformers, it's looking for ways to spread that risk. The latest idea, reports the Financial Times: insuring loans to startup cloud computing companies, or “neoclouds”, in case they default and the Nvidia chips pledged against their debt can’t be resold for enough to pay back their lenders (for example, Nvidia!). CEO Jensen Huang says he wants chips to be thought of as a new asset class. Meanwhile, Nvidia is sitting on a huge pile of cash and it's decided the best way to return that to investors (including opinion-holding managers and directors) is through the world’s largest ever stock buyback, now expected to top $235 billion. “Our cash generation gives us the capacity to invest in the technologies that advance this transformation and return capital to shareholders,” Huang said in a statement. Meanwhile, China’s state-backed AI empire is taking aim at Nvidia with DeepSeek saying it’s developed software that can run on AI chips made by hometown hero (and alleged telecom snooper) Huawei Technologies. DeepSeek said its programming language, TileLang is simpler than Nvidia’s CUDA. If it works and can scale, TileLang could help China end its dependence on Taiwan and U.S.-made chips. All of that has prompted Wall Street investment firm KKR $KKR ( ▼ 0.07% ) to warn that rising debt among tech borrowers, and the use of off-balance-sheet financing for an expected $8 trillion in AI spending leaves investors scarily exposed “We don’t think enough people are talking about the potential volatility if AI growth slows,” KKR’s Christopher Sheldon told the FT. “This is multiples on trillions of dollars of market value. The knock-on effects across the broader markets could be very meaningful.”

The Short Stack

  • Food options: Only an eight-year-old could dream up this one: Hormel Foods $HRL ( ▲ 0.1% ) , the maker of Spam along with Skippy peanut butter, is buying a chicken processing firm called Brakebush in a bid to strengthen its place with food service operators. The possibilities are stomach-churning. Hormel shares have lost half their value in the past five years as consumers search for healthier foods and stop eating so much thanks to GLP-1 drugs.

  • Man City Financial Scams: The U.K. soccer Premier League said Manchester City, the team owned 80% by Abu Dhabi’s Sheikh Mansour bin Zayed Al Nahyan, and 18% by U.S. private equity firm Silver Lake, was found guilty of financial machinations aimed at skirting league rules on player compensation. Man City “arranged ‘sham’ commercial deals” from 2009 to 2018,  in order to “artificially inflate the club’s revenues, and reduce its costs, by more than £900 million ($1.2 billion) … to appear to comply with financial rules,” the league said. Most of the money came from Abu Dhabi and appeared to be designed to let Man City buy some of the game’s best players. It recently has a roster worth over $1 billion a year. No penalty has been announced. The team said it was “innocent.”

  • Winding back Brexit? In a  bid to revive his country’s moribund economy (and shore up his party’s position ahead of mandated parliamentary elections in 2029,  Britain's new Labour prime minister, Andy Burnham, says he is looking at whether Britain could fully re-join the European Union, which it voted to leave in 2016. “We could go all the way,” Burnham said at a Labour party conference. Britain has given up six to eight percentage points of GDP growth because of Brexit, research shows, and investment has fallen by 12% to 18%. The news comes as Canada is discussing a form of associate membership with the E.U. to wean itself from depending on U.S. markets amid a trade war with the Trump Administration. 

  • YouBet: Another online betting platform wrapped in a commodities contract has seen its valuation quadruple to $2 billion after unleashing a series of racy ads featuring 29-year old White Lotus actor Sydney Sweeney. The so-called “sports prediction market” firm is called Novig, and investors include venture-capital firms Pantera Capital, Multicoin Capital and NFX. Sweeney is also reportedly an equity partner.

Trumplandia

  • Mind the gap! Forget the gap between rich and poor. Now, feel some pity for the “not quite as rich.” The top 0.1% of Americans have seen their wealth climb by more than 110% since the end of 2019, according to a Wall Street Journal analysis of Fed data, taking in $14.5 trillion in seven years, while the rest of the top 10% only saw their accumulated wealth rise by 60%. Surprisingly, the bottom 50% of Americans saw their wealth rise by nearly 130%, but their total share of the national pie only climbed to $4 trillion, versus $28 trillion for the créme de la créme. 

  • Tariff mania: The spat with Canada keeps growing, with an outright ban on imports of Canadian whiskey, motor cycles and some dairy products. Sure, drinking and driving don’t mix, even in a milkshake, but the ban comes on top of 50% tariffs on about 5% of Canadian goods last month, which saw reciprocal taxes slapped on U.S. imports North. Meanwhile, President Trump was all smiles when Chinese supremo Xi Jinping arrived in Washington, meeting Xi at the airport, agreeing to think about slashing tariffs, and, according to U.S. Ambassador to China David Perdue, offering to sell China U.S. military hardware. 

Oura’s Market Aura

(Oura)

Wearable medical device maker Oura $OURAZZX ( ▲ 0.22% ) said it’s postponing its $2.2 billion IPO, which would have valued the company at as much as $15 billion because of what it called “uncertainty in the IPO market.” Oura’s main product, a hi-tech ring that monitors vital data, sells for as much as $499, with a $60 a year app subscription. BBTW spoke with John Beadle, co-founder and managing partner at Aegis Ventures, a New York health care VC fund, to understand more about why Oura pulled its IPO.

Does Oura’s move suggest the IPO market is saturated? 

The scale of recent offerings matters. Nearly $105 billion was raised through US IPOs in the second quarter, with SpaceX $SPCX ( ▼ 1.85% ) accounting for $75 billion. Investors still have limited capital and attention, which can make it harder for other companies competing for the same buyers. I don’t think Oura’s postponement tells us demand for wearables has dried up. Oura says it has 5.7 million paid members and expects 90% revenue growth this fiscal year. Before calling a broader recovery, I’d want to see more companies successfully price offerings and perform well afterward, including those outside the biggest AI names.

Was pricing the issue?

My read is that this is a business-model and valuation question made harder by market conditions. Roughly 80% of Oura’s revenue in the nine months through June came from hardware and 20% from membership, very similar to Peloton’s mix when it went public $PTON ( ▼ 0.4% ) . Oura was profitable over that period, and membership gross margin was 89%. About 73% of shares in the proposed offering were being sold by existing holders. There’s nothing unusual about investors wanting liquidity. But a price that represents an excellent exit for an early investor still has to offer an attractive entry point for the next buyer.

How much pressure does competition from Apple and Google put on Oura?

A major challenge for Oura is that Apple $AAPL ( ▼ 0.84% ) and Google have different reasons for selling someone a wearable. The device makes the rest of their ecosystem more valuable, which gives them flexibility on pricing. Fitbit and Apple Watch offer useful core health tracking without requiring a subscription. That puts pressure on Oura to demonstrate why its ring and membership are worth paying for separately.

Are investors losing their appetite for risk amid tariffs, inflation and higher energy prices?

I think there’s a flight to quality. Investors still have an appetite for risk, but they want more confidence in earnings and a price that justifies the uncertainty. For Oura, tariffs can pressure hardware margins, while energy prices and inflation can make consumers less willing to spend on a premium ring and subscription. But I’d still put more weight on its business model and valuation.

What does this mean for companies waiting to go public?

Many companies still carry private valuations set when capital was cheaper and growth expectations were higher. That can leave a substantial gap between what their owners expected and what buyers will pay today. Bain’s 2026 report identified about 32,000 unsold companies held by private equity buyout funds, worth $3.8 trillion. Those aren’t all technology companies or potential IPOs, but they show the scale of capital waiting to be returned.

What would make Oura a durable clinical intelligence business?

First, you need clinical evidence: Does it work reliably, with an acceptable rate of false alarms and enough time to intervene? You also need the appropriate regulatory pathway. Then comes the workflow: who receives the alert, and what do they do? Finally, can you demonstrate better outcomes or lower costs that justify payment from a patient, employer or health plan? It’s a difficult path, but clearing those hurdles could support a durable clinical business.

(This interview was edited and condensed for clarity and brevity. 

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.

This post may contain affiliate links, meaning when you click the links and make a purchase, Cheddar receives a commission at no extra cost to you.