TheChek News Flow

• AI startups may struggle to hold on to valuations based on revenue
• Anthropic, others raised at a valuation of 50-100 times forward revenue
• Major cloud providers will influence AI startups’ margins

In the past year, artificial intelligence developer Anthropic has raised billions of dollars at a more than $15 billion valuation due to its fast revenue growth and big projections for future sales. But as Anthropic and scores of rivals mature, they’ll need to prioritize profit margins and cash flows if they want to build a sustainable business and have any hope of going public someday.

New data show that profit margins for such startups may end up lower than for existing enterprise software firms. After paying the costs of customer support and servers to power its AI, Anthropic’s gross margin—gross profit as a percentage of revenue—was between 50% and 55% in December, according to two people with direct knowledge of the figures. That’s far lower than the average gross margin of 77% for cloud software stocks, according to Meritech Capital.

And it may not improve much over time: At least one major Anthropic shareholder expects the company’s long-term gross margin to be around 60%, one of these people said.

Notably, Anthropic’s gross margin doesn’t reflect the server costs of training AI models, which Anthropic includes in its research and development expenses. These costs can add up to as much as $100 million per model, according to Sam Altman, CEO of OpenAI. An Anthropic spokesperson declined to comment.

The data suggest that the current breed of hot AI startup could face challenges in attracting future capital at the same prices that existing investors paid for shares as a multiple of the companies’ revenue. As companies get older, their valuation typically reflects their ability to generate cash. Right now, though, investors are focusing more on revenue growth for Anthropic, OpenAI and other young AI startups developing or enabling large language models. (See our Generative AI Database.)

These startups have been raising capital at a valuation of 50 to 100 times next year’s revenue, often due to their upbeat revenue projections. The median for publicly traded software companies, by contrast, is 6 times forward revenue, according to Meritech Capital. Those revenue multiples will be increasingly difficult for the startups to maintain as they get older.

Anthropic’s biggest startup rival, OpenAI, may have even weaker gross margins because it runs a heavily used free tier of its flagship chatbot, ChatGPT. That drives up server costs without generating additional revenue, although the free version likely helps OpenAI sell more monthly subscriptions than it would otherwise. After generating less than $30 million in revenue in 2022, OpenAI surpassed an annualized revenue pace of $1.6 billion as of December—an eye-popping growth rate. An OpenAI spokesperson declined to comment.

Anthropic, meanwhile, recently projected annualized revenue of at least $850 million by the end of 2024, after previously telling investors it would be generating $500 million in annualized revenue by then. It isn’t clear how much revenue Anthropic booked in 2023, the first year it began to generate revenue. It also isn’t clear how much money OpenAI and Anthropic are losing overall.

Rubber Meets Road

As long as the AI startups keep up that sort of momentum, investors can overlook their losses. That will likely end when revenue growth inevitably slips to 30% to 40% someday, said Tomasz Tunguz, a general partner at Theory Ventures, whose prior software investments have included Looker and Kustomer.

At that point, if a company’s cash flow from operations is negative with no near-term path to convert at least 10% of its revenue into cash flows, it’ll likely be harder to attract new investors, Tunguz said.

The growth and profit margin prospects of these startups rely in part on the major cloud providers that power their servers. Google and Amazon, which have each committed billions of dollars to Anthropic, sell its software to their cloud customers. It isn’t clear what kind of cut those cloud providers take when making such sales, but Anthropic probably generates better margins from selling its models directly to customers.

The prices the two cloud providers charge Anthropic for renting their cloud servers also have an impact on the startup’s margins. These numbers couldn’t be learned.

Microsoft’s investment in OpenAI offers a clue. As a cloud provider, Microsoft rents cloud servers to OpenAI at a relatively low margin, compared to other Microsoft cloud customers, said a person with knowledge of the situation. That helps keep down OpenAI’s server costs. As part of their deal, though, OpenAI must give Microsoft a cut of its direct sales to customers. And when Microsoft sells OpenAI software to its own cloud customers, it keeps a majority of revenue from those sales for itself. That could potentially become a problem as Microsoft increasingly draws enterprise customers that would otherwise buy software directly from OpenAI. The trend is likely eating into the startup’s growth rate.

For AI developers, the billion-dollar question is how quickly computing costs will continue to fall. Already, AI firms such as OpenAI have said they’ve cut the cost of running their AI models substantially over the past year by implementing new techniques for structuring the models, for instance. AI developers may also find it easier to run their models on cheaper servers rather than pay a pretty penny for state-of-the-art Nvidia servers. At the same time, it’s hard to predict such cost drops, as emerging models that are more computationally efficient could replace OpenAI’s LLMs.