Two years ago, during an appearance at Harvard, Sam Altman, OpenAI’s chief executive, said he hated the idea of showing ads inside ChatGPT.
If ChatGPT started answering queries with paid advertisements, he said, people would begin to lose trust in the company’s flagship product. “I kind of think of ads as a last resort for us for a business model,” Mr. Altman said.
This week, his company started showing ads inside ChatGPT.
As OpenAI spends tens of billions of dollars on the raw computing power needed to build and deploy artificial intelligence technologies like ChatGPT, the San Francisco start-up is scrambling to find new ways of generating revenue from these technologies and, ultimately, balancing its books. Selling ads inside its chatbot is just one of many ambitious efforts to make more money. All of them face enormous hurdles.
The financial pressure is immediate. Last year, OpenAI pulled in about $13 billion in revenue, according to a person with knowledge of the company. But over the next four years, it expects to spend about $100 billion more.
Mr. Altman and his lieutenants have had tremendous success raising money in recent years. But there are only so many places on the planet willing and able to part with the billions more that the company needs to pay for raw computing power.
One option is to go public on Wall Street. But even OpenAI executives privately admit that they have to stem the losses before that can happen. Hoping to triple its revenue this year, OpenAI must do many things that it has little or no experience doing. It had never before served ads, which could undermine the value of its chatbot or, worse, alienate users. It plans to make even more money by selling technology to businesses, even as a long list of rivals compete for the same dollars.
Google has been selling to businesses for decades. So has Microsoft. And Anthropic, a competing start-up, has been making gains in A.I. coding — perhaps the most notable segment of this nascent market.
OpenAI is also making claims about new business models that could drive customers away. The company recently said it wanted to take a cut of scientific discovery made using its A.I. tools. And though it later explained that this would affect only big pharmaceutical companies, the idea has unnerved many independent scientists who use its tech.
“OpenAI is trying to win consumers, trying to keep up with Anthropic’s coding tools, trying to build data centers, trying to raise more money. There are just so many things it is trying to keep up with,” said Brian O’Kelley, chief executive and co-founder of Scope3, an internet advertising company, who has spent two decades in the field. “Can it be really good at advertising? Can it be really good at all things it is trying to do?”
Last week, some OpenAI executives were surprised when The Wall Street Journal reported that the company was aiming to go public as soon as December, two people familiar with the company’s internal discussions said. Their main reason for concern was their belief that the company wasn’t ready.
At the end of last year, about 60 percent of OpenAI’s revenue flowed from its consumer products, while 40 percent came from business technologies.
Most of its consumer revenue was generated by subscriptions: Of the 800 million people who use ChatGPT, about 6 percent pay at least $20 a month for more advanced versions of the chatbot. The push into ads aims to generate additional revenue from the free version of ChatGPT.
Many veterans of the online advertising industry believe that A.I. chatbots like ChatGPT can ultimately produce billions of dollars a year in ad sales. But that could require years of experimentation. And as OpenAI experiments, it will face competition from Google and other seasoned advertising companies.
OpenAI has started to build an ad sales team, but that work is still in the early stages.
“OpenAI doesn’t really have a sales team,” said Mark Zagorski, chief executive of DoubleVerify, which works with Google and many other advertising companies across the industry. “They are going to have to build that infrastructure as well as the technology infrastructure needed to run an ad business.”
In May, Mr. Altman hired Fidji Simo, a longtime Facebook executive, to serve as OpenAI’s chief executive of applications, a new role overseeing all of the company’s many products. Ms. Simo was previously chief executive of Instacart, where she pushed the grocery delivery company toward a business model built on ads.
In the next months, OpenAI hired hundreds of employees away from X and Meta, Facebook’s parent company, where many of them worked on ad products.
Mr. Zagorski compared OpenAI to Netflix, which needed two years to build a viable ads business. In the meantime, Netflix outsourced much of its work to more experienced companies.
Even as OpenAI moves into advertising, it hopes to increase the share of it revenue from enterprise products — technologies for businesses, government agencies and other large organizations — to 50 percent by the end of the year.
“This is the critical issue on the minds of tech investors today,” said Karl Keirstead, an analyst with the investment bank UBS. “OpenAI has no choice to move more aggressively into enterprise software.”
Today, businesses pay OpenAI fees for Codex, which helps software developers write computer code, and tools like ChatGPT Enterprise, which is designed for general office use. Tools like these are widely used among technologists in Silicon Valley, with some people paying as much as $200 a month to use them.
But Mr. Keirstead said the average business might not want to pay such high rates for office software. And OpenAI faces mounting competition in the enterprise market, most notably from Anthropic and its code generator, Claude Code. As OpenAI struggles to accelerate revenue growth across both consumer and enterprise products, Anthropic is focused mainly on business tools.
Anthropic recently unveiled a Super Bowl advertisement poking fun at OpenAI’s efforts to bring ads to ChatGPT. “Ads are coming to A.I. But not to Claude,” the ad said.
Mr. Altman hit back with a post to X. “Anthropic serves an expensive product to rich people,” he wrote. “We are glad they do that and we are doing that too, but we also feel strongly that we need to bring AI to billions of people who can’t pay for subscriptions.”
(The New York Times sued OpenAI and Microsoft in 2023, accusing them of copyright infringement of news content related to A.I. systems. The two companies have denied those claims.)
Last month at the annual World Economic Forum in Davos, Switzerland, Sarah Friar, OpenAI’s chief financial officer, discussed another new way the company hoped to make money. She called it “value sharing.” If the company’s technologies help discover a new drug, for instance, OpenAI might take a share of the profits.
Days later, the company introduced Prism, a product designed for scientists. Many researchers wondered if OpenAI planned on taking a cut of their scientific discoveries, citing Ms. Friar’s comments.
Worried that they were aliening customers, OpenAI executives discussed how to handle this spiraling discourse, eventually deciding to blast countermessaging across social media.
In one post, Kevin Weil, who was recently named OpenAI’s head of science after a brief stint as head of product, explained that the company would not take a cut from the discoveries of individual scientists who used Prism. Other top OpenAI executives echoed the sentiment in posts to X. But Mr. Weil left the door open for partnerships with large drug companies in which OpenAI could grab a slice of their profits.
This week, during an appearance in Silicon Valley, Mr. Altman said much the same: “We may explore partnerships where we pay the costs and share in the upside.”
Tripp Mickle contributed reporting.
Cade Metz is a Times reporter who writes about artificial intelligence, driverless cars, robotics, virtual reality and other emerging areas of technology.
The post OpenAI’s Biggest Challenge Is Turning Its A.I. Into a Cash Machine appeared first on New York Times.
