New ChatGPT features on hold as Gemini chips away at OpenAI’s lead

OpenAI CEO Sam Altman told workers Monday that OpenAI has entered “code red” when it comes to improving the quality of ChatGPT, per an internal memo reviewed by The Wall Street Journal and The Information.

According to Altman, the company has more to do to improve the quality of its core product — specifically in personalization features, speed and reliability, and allowing ChatGPT to answer more questions. In order to focus on these goals, Altman reportedly says in the memo, OpenAI is delaying planned additional features like ads in ChatGPT, AI agents for shopping and other tasks, and a digital personal assistant called Pulse.

Though OpenAI cited weekly active ChatGPT users at some 700 million in September, Gemini is catching up quick: Google says Gemini’s active monthly users jumped from 450 million in July to 650 million in October. The Journal also notes OpenAI is challenged on some fronts by Claude developer Anthropic, which has been making inroads in the enterprise space.

OpenAI putting the brakes on its growth initiatives in favor of delivering a core product that works reliably is an unusually pragmatic development in the AI space. The company is committed to hundreds of billions of dollars in data-center investments in the coming years, despite having failed to generate profit to date. The AI industry as it stands today is largely supported on paper by a handful of companies investing in and lending to each other in so-called “circular deals,” and projections of profitability years in the future are far from certain.

The tenuous financial standing of large-scale AI operations puts OpenAI at a disadvantage versus Google, who is able to support its AI ambitions at least in part using revenue generated from other ventures like cloud services and web advertising. The Journal previously reported that competitor Anthropic is on track to generate profit before OpenAI — that company expects to break even for the first time in 2028. OpenAI, on the other hand, expects to lose $74 billion in 2028. According to the ChatGPT developer’s own projections, OpenAI may turn a profit starting in 2030 if it’s able to increase annual revenue by about $200 billion.

The many billions invested in AI software and infrastructure over the past few years have failed to generate profits generally. In November, Google CEO Sundar Pichai noted “elements of irrationality” in the AI investment climate, comparing aspects of the current market to the disastrous dot-com bubble of the 1990s.