I've all seen that guy that says he's going on a social media detox on social media. That's OpenAI. So OpenAI just confidentially filed their S1 and then announced it themselves. Their logic was, we expected to leak anyway, so we're just announcing it. Companies don't normally announce their confidential filings. That is in fact the entire point of a confidential filing. The news dropped yesterday, which is exactly seven days after Anthropic did the same thing, making it the first time two AI labs worth $2 trillion have filed to go public in the same calendar week. At SpaceX, which prices Thursday, you're looking at a $3.6 trillion stack of IPOs in one quarter. And for context, this is bigger than the entire GDP of France and bigger than every single economy in Africa combined. I have a whole thesis on the AI spend for another video, but this might be the moment that we look back on in 2040 over matcha and say, remember that crazy AI bubble? We do have a president, just not at this scale. Back in 2020, Airbnb, Doordash, Snowflake, Palantir, all rushed to list within weeks of each other. This is just 10x the size and compressed in two weeks. Now, here's the part that no one's explaining. A confidential S1 is not the same thing as filing for an IPO. You submit a draft, the SEC takes around 30 days to come back, and then you get to file it publicly. The company can also withdraw at any point during this time. So when OpenAI says timing is undecided, that it might be a while, it's not a hedge, it's an optionality dressed up as commitment. Honestly, from now on, I should call all my situationship a confidential S1. It's signals of commitment, timeline, and reserves right to withdraw at any moment. So why announce it all? Because Anthropic filed first. Anthropic, your quieter, younger sibling, went to enterprise software and quietly got jacked at Econox while OpenAI was, you know, out being famous. And Anthropic is winning the parts that actually matter. It is worth more, 9.65 billion versus OpenAI at 8.52. I'm gonna get a lot of heat for this. It arguably has better models. It owns the enterprise market. Eight of the Fortune 10 are clients and roughly seven of every 10 new enterprise customers pick cloud over chat. And it's expected to break even in 2028, while OpenAI is projected to, you know, not earn a dime until 2030. So how does OpenAI lose this much? Ironically, by building the product they said they will build. Their gross margin last year was 33%, meaning before they build anything new, just serving you a single answer cost them about 67 cents a dollar. But I really wanted to know what my chances of next winning the finals were. Then you pile on the money to train the next model. The researchers will get paid like NBA free agents. The sales, the overhead, and the marketing, and you land an operating margin of negative 122%. For every dollar OpenAI brings in, it actually spends about two dollars and 22 cents. For a company that raised close to 200 billion in funding to build a product, the product working perfectly is the thing leading them. To be fair, two things had to be clear for them to IPO, and it took some time. First, getting the permission from the two state governments to stop being a non-profit, and they fought it. Second, the $150 billion Musk lawsuit, which the jury threw out in May on statute of limitations, which is the legal version of being saved by the bell. Let's say I can get over all of these red flags. There is just one small wrinkle called Trump. OpenAI is in talks with the US government to donate an equity stake. Altman has been discussing it with the White House for apparently over a year now. And the structure sounds good. OpenAI donates equity to seed a public wealth fund, and it's pitched as a regular citizen getting a slice of the upside. Sounds great, right? The timing is not so subtle, though. Three days before the filing, Altman was sitting with Bernie Sanders, who, by the way, wants the government a 50% stake in these companies, plus a one-time 50% tax for the people. So on one side, you have a senator who wants half of your company and a 50% tax. Now, all of a sudden, donating a few percentage points doesn't seem like the worst idea, especially if you can do it on your own terms and make a big PR stunt about it. Trump administration has already taken states in Intel and like 10 other companies. And OpenAI would just be the biggest name on that list. So what none of these three are racing to go public because things are going great? They're racing because they need the money. And now is the time. OpenAI can't fund losses north of $74 billion privately anymore. And so the casual tweet, the confidential filing leak, the government courtship, it's all for the same move. So when that public S1 finally hits in about 30 days, that's the moment that the vibes meet the math. And then we get the real numbers of a fake press release. I'll be reading it the second it lands.