5 unexpected risk factors Anthropic is worried about

Anthropic (ANTH.PVT) filed paperwork for its highly anticipated initial public offering (IPO) on Monday. The IPO is widely expected to be the largest in history. It shows how much investors expect from the leading AI research company and how excited Wall Street is about the changes AI could bring.

Anthropic's shares will trade on the Nasdaq stock exchange under the symbol "ANT."

The documents, known as an S-1 (read the whole thing here), give us the clearest picture yet of Anthropic's business. Its financial results are fascinating: The company brought in $16 billion while losing more than $80 billion in the first half of the year. The filing also includes warnings about AI safety that sound like something from the Terminator movies.

But some of the most interesting details are in the section about risks, which lists the dangers the company faces. Securities lawyers are especially good at imagining what could go wrong. One of the risks even concerns the effects of negative media coverage about risks.

Many of the risks are obvious or standard warnings found in company filings. But here are five less expected risks Anthropic is worried about and that investors should watch closely:

Anthropic's customer base is small and sensitive to economic downturns

"A significant portion of our customer base consists of startups and technology-oriented enterprises that may be more sensitive to adverse macroeconomic conditions and may reduce or delay spending during a downturn," the company wrote.

One benefit of the AI business is that leading AI research companies can count some of the country's most ambitious technology companies among their customers. Why have lots of customers when you have the best ones? That works well when the economy is strong. But as Anthropic's risk warning explains, depending so heavily on customers from one industry could cause problems if economic conditions change.

If that were not enough, the company also noted that "A substantial portion of our revenue and accounts receivable is driven by a limited number of customers and distribution channels."

In other words, Anthropic has relatively few customers, and those customers may be quick to cut spending when times get tough.

Anthropic does not have other sources of income like Meta, Google, or Amazon

There is a lot of talk about major technology companies struggling to meet expectations for AI and facing questions about whether their huge AI investments will pay off. But it is important to remember that these companies were already enormous and profitable before ChatGPT became popular. They make huge amounts of money from advertising, online shopping, and cloud services. Anthropic, on the other hand, does not have a large, successful business outside AI. It has no Facebook or Amazon Prime. Most of its business is built around Claude.

"Because our cost structure limits our ability to rapidly adjust spending, a sustained economic downturn could disproportionately disadvantage us relative to larger, better-capitalized competitors. Such competitors may be better positioned to absorb or withstand adverse economic conditions or reductions in customer technology spending due to their greater financial resources, diversified product offerings and customer bases, and more stable revenue streams," Anthropic said in the filing.

In other words, Anthropic has fewer ways to change its business when conditions get tough. Meta, Google, and Amazon can offer AI tools alongside their other products and services. Anthropic depends on AI. If the AI boom suddenly ends or even slows down, the company cannot fall back on advertising or cloud service income to help make up the difference.

Anthropic is not interested in selling AI as a basic, interchangeable product

Anthropic is planning to go public at a time when AI tokens have never been cheaper and concerns about whether companies can earn enough money from their AI investments have worried investors.

This has brought more attention to the idea of cheap, interchangeable AI products. Think of a homeowner who only needs a drill once in a while. That person may choose an affordable Ryobi or Hart drill instead of paying more for the newest and most powerful model. In the same way, companies that do not need the most advanced AI may choose cheaper tools that do the job well enough.

Anthropic describes itself as "first and foremost a research lab" and says its "strategy prioritizes frontier research and model capability." The company appears comfortable focusing on advanced AI rather than competing mainly on price.

In fact, it is betting that "the advancing intelligence and capabilities of our models are the principal drivers of their adoption."

But the company recognizes the risks of a market where AI that is simply "good enough" may be all many customers need. It warns that "as the AI industry matures, the degree of differentiation among advanced AI models may narrow."

This leads to another risk: the business strengths Anthropic may be missing.

Sales, marketing, and user experience may matter more than AI capability

Anthropic identifies Google (GOOG) and OpenAI (OPAI.PVT) as its direct competitors. Meta (META) does not appear in that list, despite being a major player in AI.

Anthropic also spends time discussing what happens when AI products are used in the real world. A product's price, sales team, marketing, relationships with existing customers, and available tools may matter more to buyers than the abilities of even the most advanced AI model.

"If customers increasingly select AI providers on the basis of application features, user experience, distribution, integration into products they already use, or the strength of a sales organization rather than model capability, competitors that have invested more heavily than we have in those areas could be better positioned than we are."

This could be good news for software companies such as Salesforce (CRM), which have been hurt by concerns about AI disrupting the software business, sometimes called the "SaaSpocalypse." The same argument could also apply to the reputation-related risks described elsewhere in Anthropic's filing. Apple (AAPL) is another company that comes to mind.

When Meta introduced its new AI agent, Muse, early users praised what it could do, and analysts applauded CEO Mark Zuckerberg's efforts. But a common response to the launch was something like, "I'll wait for Google or Apple's agent."

That comes down to brand recognition and trust. Apple and Google already manage people's wallets, calendars, and emails. Meta does not. As a newer company in the AI market, Anthropic faces a similar challenge when trying to attract customers.

Many Anthropic customers can end their contracts or reduce their use of its services with little warning

The AI market is different from many other software businesses because customers can switch providers easily.

"Even if we successfully attract new customers, we face significant challenges in retaining them. Many of our customer contracts allow customers to reduce usage or terminate with limited or no notice. Customers are not obligated to renew their subscriptions or contracts with us," Anthropic wrote.

If only the rest of us could cancel gym memberships and phone plans so easily!

Customers may leave Anthropic for cheaper alternatives, build their own AI systems, or decide to switch after negative publicity. This makes it easy for them to walk away. Recurring revenue is valuable because it helps a company predict how much money it will earn next month or next year. But Anthropic's customers could leave much more quickly than the company would like.

Anthropic also warns that customers may use its services less, even if AI remains important to their businesses. That means the company could keep all its customers but still earn less money if they reduce their use of its models. Because Anthropic charges based on usage, its success depends on getting customers to use its AI more and bring it into more parts of their organizations. Business budgets can change quickly. A difficult quarter or two could lead managers to cut AI spending, whether or not Anthropic delivers on its promises.