Earlier this year, a group of OpenAI employees got together in a private chat to discuss the latest thing Sam Altman, CEO of OpenAI, posted on social media. It wasn’t the first time they had done this.
The chat, which two people said they jokingly referred to as “Making Sam’s Tweets Reality,” nodded to their boss’s penchant for announcing OpenAI plans on social media, where employees often learned about his ideas at the same time as the rest of the world. It was where they discussed how to create the surprising products he wrote about. At Mr. Altman’s behest, OpenAI has spent years tracking all AI projects, backed by billions in investment and its reputation for years ahead of technology.
Now things have changed, with competition intensifying from rivals such as Anthropic, Google and even Elon Musk’s SpaceX. Mr. Altman felt the heat — and decided to change strategy.
In the past few months, OpenAI has scrapped projects it considers “side tasks” — a term describing nonessential tasks in a role-playing game — including its Sora video generator. Instead, it has doubled down on money-making efforts, such as coding tools that can be sold to businesses. The goal, insiders say, is to be more disciplined in the face of intensifying competition as OpenAI prepares for a potential blockbuster initial public offering as early as this year.
The moves test the leadership of Mr. Altman, 41, who has faced growing criticism over OpenAI’s direction and his volatile management style.
“The same company that is trying to push the boundaries of artificial intelligence is also being asked to demonstrate financial discipline,” said Brad Gastwirth, global head of research at Circular Technology, a market research firm. “It’s not an easy balance.
OpenAI remains a leading AI lab, having raised more than $122 billion in the past year, is valued at $852 billion, and has a number of deals with industry giants such as Amazon and Nvidia. Its revenue has grown to an “annual rate” of $24 billion, a long-term revenue estimate based on short-term data, and its ChatGPT app is regularly used by more than 900 million people.
But the others caught up. Google has placed AI-based features at the forefront of its widely used products. SpaceX, which aims to build AI data centers that “orbit the Earth,” said this week that it is working with artificial intelligence startup Cursor to improve its AI models.
Perhaps most threatening is Anthropic, a San Francisco artificial intelligence company that has built a thriving business selling coding tools to individuals and businesses. This month, Anthropic he said its annual revenues have surpassed $30 billion, up from $9 billion by the end of 2025. It’s launching new AI products almost every week that have shaken up long-standing software companies.
An OpenAI spokesman declined to comment or make Mr. Altman available for an interview. The CEO may be busy with other matters before a trial begins next week in federal court in Oakland, California, over the 2024 lawsuit Mr. Musk has filed against him. Mr. Musk argued that Mr. Altman violated OpenAI’s founding agreement by prioritizing commercial interests with AI over the public good. OpenAI has denied the allegations.
“We want to be a platform for every company, scientist, entrepreneur and person,” Mr. Altman said on the social network. post on Thursday announced the latest OpenAI artificial intelligence model. “We want our users to have access to the best technology and for everyone to have equal opportunities.”
(The New York Times sued OpenAI and its partner, Microsoft, accusing them of infringing news content related to AI systems. OpenAI and Microsoft have denied the claims.)
When OpenAI released ChatGPT in 2022, the chatbot quickly topped the download charts and broke growth records. The momentum gave Mr. Altman leverage to attract billions of dollars in investment and attract top AI talent.
Six current and former OpenAI employees said it also lulled him into a false sense of security that allowed for distraction. Friends and colleagues have described Mr. Altman as someone who enjoys chasing moonshot-like products, even if they prove expensive or impractical.
OpenAI soon jumped into many eclectic projects. One team worked to create an “unsafe for work” chatbot that could engage in violent conversations with users. Another team created a separate social network where people could create AI videos and share them over the internet. Yet another team was tasked with developing ways to speed up AI scientific research.
Then, over the past 18 months, OpenAI’s lead evaporated as Google, Anthropic and others gained momentum.
“The environment is constantly changing so quickly and so unexpectedly that the mood of different companies can change practically overnight,” said Rayan Krishnan, CEO of Vals AI, an AI start-up.
At the end of last year, OpenAI managers took action. The company has hired Fidji Simo, CEO of Instacart and former vice president of Facebook, to become its CEO of AGI for deploying artificial general intelligence. In practice, Ms. Simo became the company’s day-to-day operator and helped Mr. Altman streamline the strategy, three current and former employees said.
“We really need to improve productivity,” Ms. Simo said in internal memos to staff that were provided to The Times. She stressed the importance of catching up in areas dominated by competitors such as Anthropic, characterizing the situation as a “code red”.
In recent months, Mr. Altman, Ms. Simo and Sarah Friar, OpenAI’s chief financial officer, have reviewed the company’s projects to see what works and what doesn’t, the three people said.
They decided to invest in Codex, OpenAI’s enterprise coding product, with the idea that they could sell coding tools to businesses while simplifying parts of ChatGPT to create what Ms. Simo characterized as a “super app” for desktops. They also create advertising products at ChatGPT. Formerly the Wall Street Journal reported about the application plans.
Not everyone is joining the next phase of OpenAI. Three executives announced their departure last week after Mr. Altman canceled or reorganized parts of the company, including some scientific initiatives and Sora. Some senior executives have expressed reservations about whether OpenAI will be ready for an IPO this year, two people familiar with the internal discussions said.
Last month, OpenAI gave employees a week off to avoid burnout, two people familiar with the matter said. Other current and former employees characterized the time off differently: the calm before the storm.
