OpenAI board voices doubts about feasibility of $280B-by-2030 revenue goal

AI revenue is the easy part of the story. AI costs are the hard part. That is why this OpenAI fight matters. The company has been pitching a future where ChatGPT, enterprise software, and developer tools grow into a business doing more than $280 billion a year by 2030. But inside the company, the board and finance team are pushing on a simpler question — can that growth arrive fast enough to pay for the machines first? (cnbc.com) ### What actually changed? The new thing is not just the $280 billion target. OpenAI also reset how it talks about spending. In February, it told investors it was aiming for roughly $600 billion of total compute spend by 2030, down from a much louder $1.4 trillion infrastructure figure Sam Altman had floated earlier. That was already a sign that investors wanted a more believable story linking spending to revenue. (cnbc.com) ### Why is the board uneasy now? Because the revenue ramp has looked less automatic than the hype suggested. In late April, reports said OpenAI had recently missed internal goals for user growth and revenue. CFO Sarah Friar reportedly warned other leaders that if growth does not accelerate, OpenAI could struggle to cover future computing contracts. Separate repo(cnbc.com) locking in more capacity while growth has cooled. (money.usnews.com) ### Why does $280 billion sound so aggressive? Because the jump is enormous even by AI standards. OpenAI generated $13.1 billion of revenue in 2025, then told investors it expects more than $280 billion in 2030. That implies a business that has to keep compounding at a ferocious pace for years while competition gets t(money.usnews.com) immediately.” (cnbc.com) ### Where is that revenue supposed to come from? From two big buckets — consumer and enterprise. OpenAI told investors those would contribute roughly equally by 2030. Consumer means ChatGPT subscriptions and related products. Enterprise means companies paying for APIs, workplace tools, and custom deployments. The catch is that both buckets are expensive to win. (cnbc.com)ute guarantees. (cnbc.com) ### Why is compute the real choke point? Because AI is not a normal software business where extra users are almost free. Training frontier models costs a fortune up front, and serving them costs money every day after that. So OpenAI is trying to build demand and supply at the same time — sign users, sign enterprise contracts, and reserve enough chips and data-ce(cnbc.com)t looking like a mortgage taken out on a future that has not shown up yet. (cnbc.com) ### Is competition making the math worse? Yes. Reports say OpenAI lost some ground to Anthropic in coding and enterprise markets, while Google’s Gemini kept growing. That matters because those are exactly the categories where high-value customers live. If rivals slow OpenAI’s pricing power or force it to spend more to keep users, the path to that 2030 number gets narrower fast. (money.usnews.com) ### What does this mean for an IPO? It means the story has to shift from “AI leader” to “AI leader with believable unit economics.” Friar has reportedly been more cautious than Altman about rushing into public markets. That makes sense. Public investors can tolerate huge losses for a while, but they want a clean explanation of how spending converts into durable margins. Right now, the board seems to be asking for that explanation before blessing the next giant bet. (money.usnews.com) ### Bottom line? OpenAI is not backing away from scale. It is being forced to prove that scale will pay. The $280 billion target is really a referendum on whether frontier AI can become a giant business before its infrastructure bills come due. (cnbc.com)