Wall Street braces for anthropic spending

Dario Amodei, CEO of Anthropic, pitched his AI company to potential stock market investors with a lofty vision this week. He predicted the end of cancer and mental illness within ten years and the transformation of cognitive work. “If we do it right, we can create an age of miracles,” he said marketing video for Antropic’s initial public offering.

Wall Street is grappling with how much it might cost to get to utopia.

Equity analysts at Goldman Sachs, the investment bank leading Anthropic’s IPO, told investors privately this week that they forecast the company will burn through more than $65 billion next year and another $41 billion in 2028, two people familiar with the discussions said. Those sums would make Anthropic’s spending one of the company’s most aggressive bets to finance the construction of an AI infrastructure giant.

“We are in an aggressive investment mode,” Krishna Rao, Anthropic’s chief financial officer, said in the video, without discussing specific spending forecasts.

Those expenses would paint a drastically different financial picture than the one Antropicky recently showed. The company generated cash from its business in the first nine months of this year, his IPO prospectus shows performance in the expensive business of developing AI models. It barely spent the $141 billion it raised privately from investors.

A Nasdaq listing, expected next Thursday, could value Anthropic at $1.4 trillion. Not only would this cement its position as one of the most prominent companies in the world, but it would also give it additional firepower to help it pull further ahead of rivals such as OpenAI and Alphabet. An IPO would give Anthropic an additional $100 billion to spend and make it easier to raise even more money to fund these huge liabilities.

But Anthropic’s public debut will lead to costs not seen today. It said in a prospectus that it expects to spend about $40 billion in the coming years related to pre-existing stock grants it has given to employees that have not yet hit its books.

The company’s prospectus for investors put the future bill Anthropic has spent on chips, power and other computing infrastructure to help it develop more advanced artificial intelligence models at $900 billion over the next two decades.

Stock pickers and analysts said the prospectus lacked more detailed spending details, such as how much goes to AI training and research, customer service and its own internal use of its AI service Claude.

“There’s definitely a desire to focus on the huge, broad huge opportunity that’s in front of us rather than getting into the specifics,” said Ashley MacNeill, managing director of private equity firm Vista Equity Partners.

Ms. MacNeill, a former investment banker who worked on high-profile technology IPOs, added that she did not think Anthropic was “trying to hide the ball,” but said the lack of information “creates uncertainty about how they will manage this company from a cost perspective going forward.”

Regardless, investors are bracing for Anthropic to raise the bill. The size of the company’s existing financial commitments “will likely fall significantly short of what Anthropic ultimately needs,” investment bank analysts William Blair wrote to investors this week.

The report characterized Anthropic’s pace of constantly releasing new versions of Cloud as a “model treadmill”. Analysts wrote that the company’s potential valuation was attractive, but added that “we sympathize with investors who point to the company’s risk in a 4-year-old industry.”

Part of the challenge for investors is determining exactly what kind of business they are signing up to support. Anthropic’s rise has been dizzying. Within a year, the company had grown quarterly revenue from about $1 billion, about the size of Tinder dating app parent Match Group, to $17 billion, the size of more than century-old stalwarts like Pfizer and IBM.

Antropia executives emphasized that they have much more room to grow. Its focus on business customers, rather than consumers, helped it leapfrog OpenAI maker ChatGPT in sales this year. Its success was largely due to strong adoption by software developers. They are also targeting a wider range of white-collar workers to automate more diverse work tasks.

“For most workers, Claude is still just a chatbot,” Nicholas Lin, an Anthropic employee, said in an interview Monday. The company is trying to change that. On Tuesday, it introduced new tools for people to include Claude as a participant in a work meeting or whiteboard.

The company will need these products to catch on. In its private forecast for investors, Goldman Sachs predicted that Anthropic would exceed $440 billion in annual revenue by 2029, more revenue than Apple generated last year. That increase would help Anthropic generate enough cash to cover costs through this year, according to the forecast. Financial Times formerly reported on the forecast of the investment bank.

However, Anthropic’s customers are not bound by long-term contracts, and the vast majority of the company’s revenue comes from versions of its AI models it has released in the past six months, the prospectus said.

Shivaram Rajgopal, a professor of accounting and auditing at Columbia Business School, said that could be a problem if Anthropic can’t gain enough customer loyalty and keep prices high.

At worst, the company could end up with a similar business dynamic to Netflix, which must continue to spend heavily to produce shows to keep subscribers. “That’s the risk of the business model,” said Dr. Rajgopal. “They are slow moving disasters.

Antropic said the cost of computing to develop new AI models is worth it. The spending, it argued, creates a reinforcing cycle that will help it maintain its lead in the AI ​​race and ultimately generate profits. Management says customers continue to spend on Claude because its new model versions keep getting better, and the company can also reduce costs for customers and itself as the models improve.

The company sees itself as the ultimate beneficiary of its own creation, as it uses Claude to automate and accelerate the development of its models.

Anthropic published message last month, which suggested that its AI agents, or tools that can complete work tasks like writing code or creating spreadsheets, worked six times more hours per week than its human employees. “Our pace of discovery is accelerating, and a big part of that acceleration is coming from Cloud itself,” Sam McCandlish, co-founder and chief architect of Anthropic, said in a video to investors.

Many investors considering the deal didn’t see how Claude transformed his work in the same way. “You see anxiety and disconnect between AI believers and non-believers,” Ms MacNeill said. “We’ve only been sitting with this technology for a few years.”