
The Financial Times reports that OpenAI expects to lose at least $278 billion through the end of 2030. This, the publication says, is the real reason the company recently pushed back its initial public offering (IPO): It needs to achieve a higher valuation of $1.2 trillion first so that it can raise even more money to fund its aggressive infrastructure expansion.
This stunning report is based on internal OpenAI documentation and it represents the company’s real expectations for the business. And the $278 billion figure doesn’t account at all for past losses: This is the amount of money OpenAI expects to lose during the five years between 2026 and 2030.
The documentation states that OpenAI expects to lose $36 billion in 2026, but its losses will expand over each subsequent year and should hit $350 billion in losses in 2030. Revenues will likewise increase roughly tenfold, too, from $36 billion in 2026 to an estimated $350 billion in 2030. So OpenAI internally expects to generate cumulative revenues of $840 billion by the end of 2030.
But revenues are not profits. OpenAI will have a negative cash flow of $278 billion over the next five years. And it plans to spend $856 billion on AI infrastructure costs during that period in a “fake it until you make it” bid to somehow make the business viable. But it can’t do so unless it raises more money: OpenAI most recently raised $122 billion this past March, but it’s burning through that money so quickly that the cash will be gone by 2028.
OpenAI was valued at $852 billion after its March funding round. It filed for an IPO in June and it originally expected to go public before the end of 2026. But the company said a week ago that it would delay the IPO until 2027 to give it time to focus on AI safety. The leaked documentation shows otherwise. With no path to profitability, the AI spending led by OpenAI is well on its way to becoming the most expensive lie in human history.