Anthropic’s Path to $1.5 Trillion Demands 30% Margins: Plausible, but Delivering Is Another Matter - Apple…

NYU Stern finance professor Aswath “The Dean of Valuation” Damodaran warns that Anthropic's upcoming IPO, when it first sells shares to the public, will be priced based on what investors are willing to pay, rather than what the company is actually worth. He believes market excitement and investor behavior will decide the price, not the company's underlying financial value.

At an October 15 meeting before its possible IPO in San Francisco, Anthropic reportedly said it earned $16.8 billion in revenue during the third quarter. Its annualized revenue run rate (ARR), an estimate of yearly revenue based on its current pace, reached $67 billion. The company expects that figure to reach $100 billion by the end of the year.

Damodaran sees something similar to what happened in the late 1990s. Back then, internet companies were priced based on how many people visited their websites. Later, investors used a measure called “eyeballs,” which looked at how much time people spent on those sites. Today, he believes ARR plays the same role for AI companies. It gives investors another way to judge companies whose actual financial results do not seem strong enough to support their high market prices.

Anthropic Is Young but Already Huge

Using Damodaran's way of looking at the stages of a company's life, Anthropic is a young, fast-growing business that is already extremely large. Its revenue is growing very quickly, but it is losing money from its operations, spending heavily to grow, and using more cash than it brings in. Damodaran says these are “features not bugs,” meaning they are normal parts of this stage of growth rather than unexpected problems.

However, he does not accept Anthropic's claim that it made an adjusted profit in the third quarter. An adjusted profit is a profit figure calculated after leaving out certain costs.

“I don't think even Anthropic believes it made money,” he said. He also described adjusted financial figures as “often a fictional number.” The company's free cash flow, meaning the cash left after paying for its operations and major investments, is still deeply negative.

One major concern is how much money Anthropic needs to spend on computing power. The company has more than $400 billion in contractual commitments for computing resources. These are agreements that require it to pay for the technology needed to run its AI systems. Because Anthropic is still losing money and the AI industry is still taking shape, these commitments create serious risks. Investors must decide what the company could earn in the future instead of relying on a proven record of producing cash.

Why ARR Can Be Hard to Trust

Damodaran writes that ARR is “not standardized and can be easily gamed.” In simple terms, companies do not all calculate it the same way, and they can choose methods that make their numbers look better. A company might take the revenue it earns in one day, one week, or one month and use that to estimate a full year's revenue. Both leading AI companies currently use monthly revenue figures, but Damodaran warns that “there is nothing that binds them to continue to do so.”

Different accounting methods also make companies harder to compare. OpenAI changed to reporting net revenue, which lowered its reported ARR from $70 billion to $50 billion. Net revenue leaves out certain payments or costs when calculating revenue. Anthropic still reports gross revenue, meaning revenue before those deductions, and counts payments to cloud providers as expenses instead.

If Anthropic is priced at $1.5 trillion, that would equal about 23 times its current ARR. But compared with its projected revenue of $300 billion in 2030, the price would be only five times that amount.

Damodaran expects people who strongly believe in Anthropic's future to focus on those much later revenue estimates. “You're going to see a lot of Anthropic bulls use forward numbers. Really, really forward numbers,” he predicted, “because it makes your metrics look more reasonable.” In other words, using revenue estimates from several years in the future can make today's very high price seem easier to justify.

The Problem With Finding Similar Companies

One common way to decide a company's price is to compare it with similar businesses. But there are almost no publicly traded companies focused only on large language models (LLMs), the AI technology behind systems like chatbots, that can be used for a fair comparison.

One option is to compare Anthropic with OpenAI, which is also privately owned. But because neither company is publicly traded, investors must rely on prices from earlier private fundraising rounds, which may already be outdated. Another option is to compare Anthropic with the Magnificent Seven, a group of seven major technology companies, and Space Exploration Technologies Corp. (NASDAQ:SPCX). However, these companies are very different in how established they are, how they are managed, and how much cash they generate.

“Apple Inc. (NASDAQ:AAPL) is a very different company than Anthropic,” Damodaran pointed out. Investors who are doubtful about Anthropic will argue that the big differences in revenue and profits between these companies “cannot be explained away with hope and potential.”

Damodaran's main point is that every method of pricing a company using a multiple, such as comparing its price with its revenue, quietly depends on assumptions about how much cash the company will make in the future. This is the same basic idea used in a discounted cash flow model, which estimates what future cash earnings are worth today.

“When you price companies, you're making exactly the same assumptions I will have to make to value the company. You're just making them implicitly,” he said. This means investors are making predictions about the future even when they do not clearly show those predictions in their calculations.

Damodaran believes Anthropic would need to achieve several major goals to justify a value of $1.5 trillion to $2 trillion. Its revenue would need to grow extremely quickly. Its operating profit margin would need to reach at least 30%, meaning it would keep $30 or more in operating profit for every $100 in revenue. Its computing costs would also need to grow more slowly than its revenue. Finally, the company's overall level of risk would need to move closer to that of an average business as threats from government rules and regulations become less serious.

Damodaran remains careful about these predictions. He says Anthropic's possible path to such a high value is “a plausible one,” meaning it could happen. But he warns that actually achieving all those goals is a very different and much harder challenge.

Image via Shutterstock