$50 billion or $70 billion? Why OpenAI’s revenue numbers don’t add up

OpenAI is back in the news. Bloomberg reporting overnight that OpenAI expects to reach or exceed $70 billion in annualized revenue by year end. Annualized revenue is the figure that the labs such as OpenAI and Enthropic like to use, >> but they don't even use the same metric. >> They don't they So, this is part of the problem. They don't quantify it the same way. Um, OpenAI hit approximately $50 billion at the end of September in uh annualized revenue. They're expecting now to hit 70 or reach that by year end. That said, at basically the same time we got this Bloomberg report, the FT, the Financial Times comes out and says their reporting shows it's actually going to be closer to 50, right, >> than the $70 billion target. The problem here, or at least part of the problem, is how these companies do their accounting. >> Yes, >> annualized revenue is not a I mean, it's not not a real number, but it's not a figure you would typically see from a real earnings report like you just report your quarterly or your annual revenue. The reason you annualize it is to get rid of some of that lumpiness and some other reasons. But the problem here comes down to how they calculate it. Calculate it, excuse me. OpenAI recognizes only its share of revenue through cloud partners that it gets uh like it only for the deals it has with cloud partners. It only recognizes its own share of that revenue. >> Anthropic counts the uh the gross amount of certain cloud partner sales. So it's the >> they quibble with how they count what is theirs versus what is somebody else's. And so that leaves you in a position where they could be doing the same amount of business essentially, but then arriving at two vastly different figures. 50 billion versus none of this matters. The S1's are ready. So the none of this matters until they IPO. >> That's the thing. Like, oh god, all these all these numbers are fake. Drip drip drips. like let's just get the S1 so we have the actual numbers and so we can actually get our arms around what these businesses are and you know I'm tired of all the conjecture I know analysts have to have a view I know it is important for the market to have some sense of if these companies are making money but until we get both S1's all conjecture because these are private companies who do not actually have to report anything >> right >> so is it weird then we're seeing all these the the names associated to the AI trade like getting brocked by these headlines right Obviously, it makes sense that they are, but like you said, nothing's confirmed. It's all backward-looking data. We don't know what's new. >> Here's the problem. In the absence of certainty and in the absence of clarity, the market seizes on this this these leaks and this grasp onto what you have grip of of information. And it's unclear if the opening the the FT report yesterday where they talked about this $20 billion sort of understatement, is that really what was going on? Was it just a misunderstanding of what they were measuring and on for what time period? It's unclear to me exactly what's going on there. But yeah, it put pressure on all these stocks yesterday and it looks like they're that sell off continuing. >> It's also it's also worth just pointing out the way you calculate annualized revenue or the way they're seem to be doing it is by taking some short window of the revenue you made revenue you made, excuse me, and then extrapolating that out for the rest of the year. >> Yeah. which anyone would tell you is kind of an assumption on on several different frames. >> Like if you assume that something's going to going to double over and over and over again, you get to unrealistic numbers. >> Yeah. Also, you could look look at look at Apple, right? You take their holiday quarter and like they're annualizing revenue the holiday quarter this this it's like >> right and well and the and the I think our assumption is is that they are doing it for that reason because it is lumpy. They want to smooth it out >> and growing and growing. Right. >> Right. and growing. Whereas maybe it's sort of moving in fits and starts as you win different contracts. >> Good projections of annualized revenue are good for valuation ahead of an IPO. Opening eyes is raising another $30 billion round to get them valued at 1.4 trillion. The way you do that is by telling your VCs, your funders, all of those people, look how fast our revenue is growing. Look at where we're going to be. That's how you justify we should be worth$1.5 trillion dollars. I wonder if you know with SpaceX unique sort of lockups schedule for the insiders. Are we going to see that with anthropic? Are we going to see with open AI? I don't know because it looks like we have two different sides, right? We have the sides of people were very skeptical in the market with the regards to the pressure felt by the core reeves of the world, right? And then you have the other side of the the private fundraising for OpenAI. >> Yes, it's not as much as they wanted, but it's still go a lot and same with anthropic. Are these guys just want to get in now because they can want to they can get out later at a lot higher value? Is that what this is all just a trade or is it true believers? >> Right. >> It's a great question. >> This this is also >> and with that >> revenue is only one half of the picture. >> Yeah. >> The other half of the picture is all the debt getting issued by this industry. And so when we get those S1's the really critical number is going to be how much money are you actually making in reality with hard numbers because X amount of hundreds of billions of dollars of debt is now on the on the credit market at some point as we keep saying every single day. Someone's got to pay that off someday. Someone's got to hold the bag. >> In the auto world, right, to build a factory costs and like a decentsized factory, >> $5 billion. It's a lot of money. People freaked out. It's so much spending. It's so much. It's huge commitment. $5 billion in this world. It's nothing. It's like a closet in a data. It's nothing. Exactly. You know, it's it the numbers mean nothing. >> Yeah. We all We all scoffed at a company that was valued at something like $200 million. It's like you're not even playing in the same ballpark. You're in the you're in the little league. He softened.