Sam Altman Back Tracks Mind-Melting $1.4 Trillion Claim With $800B Haircut…

“You sit on a throne of lies” - Jensen Huang, probably

About five months ago, Sam Altman had every OpenAI seed investor pounding nose beers after he stood in front of the world and said OpenAI was going to spend $1.4 trillion on infrastructure (read: Nvidia chips) by 2030. Fast forward today, and it appears @sama lied out of his a$$. 

(Source: Giphy) 

In short, out goes the $1.4 trillion number and in comes the new $600 billion number. For those keeping score at home, that’s an $800 billion haircut in the name of premature excitement. Why? Well, let’s run it back. If you recall, OpenAI went on an absolute spending spree in the back half of last year… inking multibillion-dollar infra deals with every chipmaker and cloud provider it could find. Altman was out there making $1.4T sound like a down payment. The energy was electric… the bean counters, apparently, were not.

Now the company is telling investors the compute target is "roughly $600 billion" through 2030, and that this number is meant to "more directly tie to expected revenue growth." Which is a very polite way of saying somebody finally did some math and said “oh no”. As for the revenue growth claim…  OpenAI is projecting $280 billion in 2030 revenue, split roughly evenly between consumer and enterprise. For context, they did $13.1 billion in 2025 (beat their own $10B target, credit where it's due). So they need to 21x revenue in five years. Woof.

The burn rate, at least, came in slightly better than expected…  $8 billion scorched versus a $9 billion target. So they're efficiently lighting money on fire. Progress. Meanwhile, the fundraising machine keeps humming. OpenAI is closing a round that could top $100 billion, with about 90% coming from strategic investors. Nvidia is reportedly in talks to drop up to $30 billion into the round at a $730 billion pre-money valuation. SoftBank and Amazon are also in the mix. So the company that just told you it's spending $800 billion less than planned is simultaneously raising the GDP of a mid-sized country. Sounds legit. 

(Source: Giphy) 

That said, ChatGPT is back to record highs with 900 million weekly active users, up from 800 million in October after a brief autumn dip that apparently triggered a full "code red" internally. Additionally, Sam’s coding tool Codex crossed 1.5 million weekly actives and is going head-to-head with Anthropic's Claude Code, which has been quietly eating its lunch in developer circles. 

Now of course, will the $600B number be enough to seal the AI throne? Who knows. It’s still an absolutely psychotic amount of money to spend on compute. But at the end of the day, when you anchor at $1.4 trillion, $600 billion sounds like fiscal discipline. And maybe that's the whole point. Maybe the $1.4T number was never real. Maybe it was always a fugazi. Aaaaaand if you listen closely, you can hear investors wondering, “wait a second, will this thing ever make money?” Until next time, friends… 

At the time of publishing, Stocks.News holds positions in Amazon as mentioned in the article.