With a demand for hundreds of billions in funding, OpenAI to launch ads on ChatGPT this week

Two years ago, Sam Altman, while attending an event at Harvard University, stated that he was strongly opposed to the idea of placing ads in ChatGPT.

He said that if ChatGPT started inserting paid advertisements into its responses, users would gradually lose trust in the company’s flagship product. “I think advertising is the last resort for us as a business model,” Altman said at the time.

And yet, this week, his company began displaying ads in ChatGPT.

To build and deploy AI technologies like ChatGPT, OpenAI needs to invest tens of billions of dollars in foundational computing power. The San Francisco-based startup is urgently seeking new ways to generate revenue from these technologies in order to ultimately break even. Selling ads in its chatbot is just one of many revenue-increasing plans, all of which face significant obstacles.

The financial pressure is mounting. According to people familiar with the matter, OpenAI generated about $13 billion in revenue last year. But over the next four years, the company is expected to invest another $100 billion.

In recent years, Altman and his team have achieved great success in fundraising. But globally, investors who are willing and able to keep coming up with billions to support its computing costs are few and far between.

One option is to go public on Wall Street. But even OpenAI executives privately admit that the company must stop its losses before listing. The company hopes its revenue will triple this year, which means it must venture into many areas it has little to no experience in.

It has never run an advertising business before, something that could diminish the value of its chatbot or, worse, alienate users. The company also plans to generate more revenue by selling its technology to enterprises, but meanwhile a long list of competitors are fighting for the same market.

Google has been providing services to enterprises for decades, and so has Microsoft. Meanwhile, upstart competitor Anthropic has been making significant inroads in the AI programming space—perhaps the most watched segment of this emerging market.

OpenAI is also promoting new business models that could potentially drive away customers. The company recently said it hopes to take a share of scientific breakthroughs generated using its AI tools. Although it later clarified that this only applies to large pharmaceutical companies, the idea still unsettled many independent scientists using its technology.

“OpenAI is trying to win over consumers, catch up with Anthropic’s programming tools, build data centers, and keep fundraising all at once. There’s just too much it’s chasing,” said Brian O’Kelley, CEO and co-founder of internet advertising company Scope3, who has twenty years of industry experience. “Can it really do advertising well? Can it really do everything it wants to do well?”

Two people familiar with internal company discussions said that when reports surfaced last week suggesting OpenAI might go public as soon as December, some executives were surprised. Their main concern was that they believed the company was not ready.

At the end of last year, about 60% of OpenAI’s revenue came from consumer products, and 40% from enterprise technology.

The majority of consumer revenue comes from subscriptions: among the 800 million ChatGPT users, about 6% pay at least $20 a month for access to more advanced versions. Expanding into advertising aims to generate additional revenue from the free version of ChatGPT.

Many veterans of the online advertising industry believe that AI chatbots like ChatGPT could eventually generate billions of dollars in annual ad revenue. But this may require years of experimentation. And as OpenAI explores, it will also face competition from established ad giants like Google.

OpenAI has begun assembling an ad sales team, but this work is still in its early stages.

“OpenAI doesn’t really have a true sales team,” said Mark Zagorski, CEO of DoubleVerify, which works with many ad companies including Google. “They need to build the infrastructure and technical systems required to operate an ad business.”

In May of this year, Altman hired Fidji Simo, who had a long tenure at Facebook, as CEO of OpenAI’s applications business, a new position overseeing all company products. Simo previously served as CEO of Instacart, where she led the grocery delivery company’s shift to an ad-centric business model.

In the following months, OpenAI poached hundreds of employees from X and Meta (Facebook’s parent company), many of whom previously worked on ad products.

Zagorski compared OpenAI to Netflix, which took two years to build a viable ad business. During that time, Netflix outsourced most of the work to more experienced companies.

Even as it enters advertising, OpenAI still hopes to increase the share of revenue from its enterprise products—technologies aimed at businesses, government agencies, and other large organizations—to 50% by the end of this year.

“This is the key issue that tech investors care about today,” said UBS analyst Karl Keirstead. “OpenAI has no choice but to push more aggressively into the enterprise software market.”

Currently, enterprises pay OpenAI for tools like Codex (which helps developers write code) and ChatGPT Enterprise. These tools are widely used in Silicon Valley’s tech scene, with some users paying as much as $200 per month.

But Keirstead noted that ordinary enterprises may not be willing to pay such high fees for office software. And OpenAI faces increasingly fierce competition in the enterprise market, especially from Anthropic and its code generation tool, ClaudeCode.

While OpenAI is striving to accelerate revenue growth on both the consumer and enterprise fronts, Anthropic is mainly focused on enterprise tools.

Anthropic recently aired a Super Bowl ad, poking fun at OpenAI’s move to introduce ads. “The age of AI ads has arrived—but Claude has no ads,” the ad said.

Altman fired back on X: “Anthropic sells expensive products to rich people. We’re glad they do that; we do it too, but we also strongly believe we need to bring AI to the billions who can’t afford a subscription.”

At last month’s World Economic Forum annual meeting in Davos, Switzerland, OpenAI CFO Sarah Friar discussed another new approach to profitability for the company. She called it “value sharing.” For example, if the company’s technology helps discover new drugs, OpenAI may share in some of the profits.

Days later, the company launched Prism, a product for scientists. Many researchers cited Friar’s remarks, questioning whether OpenAI intended to take a cut of their scientific discoveries.

Concerned about alienating customers, OpenAI executives discussed how to respond to the growing public debate and ultimately decided to clarify the situation on social media.

In a post, Kevin Weil, who had just been appointed as OpenAI’s Chief Science Officer, explained that the company would not take a cut from individual scientists using Prism. Other executives expressed the same stance on X. But Weil did not rule out the possibility of cooperating with large pharmaceutical companies and OpenAI sharing in the profits.

This week, Altman expressed a similar view while attending an event in Silicon Valley: “We may explore some partnership models where we bear the costs and share in the proceeds.”

Editor: Chen Yujia