OpenAI Leaders At Odds Over IPO Plans

According to a report Sunday (April 5) by The Information, CEO Sam Altman has said he wants the IPO to happen as soon as the fourth quarter, even though the artificial intelligence (AI) startup will spend upwards of $200 billion before it begins to generate cash.

Meanwhile, Chief Financial Officer Sarah Friar has suggested in-house that the company may not be ready to go public this year, a source who spoke with her told the news outlet. This source said that Friar has also voiced concerns about OpenAI’s financial exposure related to steep spending on computing infrastructure. 

The report also notes some tension behind the scenes at OpenAI, with sources saying that Altman had excluded Friar from conversations with investors and from meetings that involved important financial decisions. 

Last year, the report added, Friar began reporting to Fidji Simo, the former Instacart chief executive who had been named the CEO of OpenAI’s applications business.

The Information report includes a statement from both executives, which says: 

“We are fully aligned that durable access to compute is at the core of OpenAI’s strategy and a key differentiator as we scale. We have both been directly involved in every consequential compute decision over the past year plus.” 

The statement added that OpenAI’s recent $122 billion funding round “locks in the capacity to scale compute aggressively and positions us to become the core infrastructure layer for AI.”

Meanwhile, a report Sunday by the Wall Street Journal (WSJ) examines the “Achilles heel” facing both OpenAI and rival Anthropic as they prepare to go public: the high cost of training new AI models.

According to that report, OpenAI anticipates it will spend $121 billion on computing power for AI research in 2028. This means the company expects to burn through $85 billion that year even after nearly doubling sales from the previous year. Losses of this scale, the WSJ added, would surpass those of almost any public company on record.

Training costs are so high, the report said, that both startups use two measures of profitability: one with training costs, one without.

Using the latter metric, OpenAI is on pace to realize a small pre-tax operating profit this year. Using the former, OpenAI won’t break even until the 2030s, the report said.

An OpenAI spokesperson told the WSJ the company prioritizes growth over profits, and could reduce spending on training but expects a robust return on investment.