OpenAI has reportedly filed confidentially to go public, and now we’re back at the part nobody can dodge: when the nonprofit says stop and the shareholders say go, who actually wins? This is Musk v Altman Daily, and today the IPO question is no longer hypothetical, because there’s apparently an actual document sitting at the SEC right now. John Polonis is naming the dual-loyalty problem in the S-1, and Karen Hao — she was physically inside OpenAI — is putting a human face on what that race looked like from the inside. She called it a race to “Great God” — and she was in the room. That’s not some outside critic reaching; that’s a witness talking. Technocracy News, with Karen Hao:
For the last three weeks the world has watched as two of the world’s richest men, Elon Musk and Sam Altman, engaged in a public mudslinging battle through a California court about an organisation they co-founded: OpenAI. The evidence provided, including memos, emails and text messages, all gave a tantalisingly rare window into the origins of the company.
The Technocracy News piece from May 27th is worth a careful look. It’s a republished interview, but the source is Karen Hao — she had physical access to OpenAI’s offices in 2019, and she’s since talked to hundreds of former employees. So no, not a deposition, not an investor memo. Just a reporter who was there saying the trial made her feel vindicated. Even the book title — Empire of AI: Inside the reckless race for total domination — tells you where she’s coming from. And that phrase “Race to Great God” didn’t come from some outsider tossing rocks. It came from someone who was inside OpenAI when the nonprofit-to-profit pivot was happening. And legally, that matters. The trial never decided the founding-mission question — the jury went to timing, not whether the mission was ever real. So Hao’s account is probably the closest thing to an eyewitness record of what that internal culture looked like during the exact period the court was looking at. Meanwhile, the Wall Street Journal is saying a trillion-dollar IPO is in motion, so now the “race to great god” framing and the S-1 are headed straight toward each other. Devin raised this on Tuesday: if the nonprofit is still nominally in charge after restructuring, what happens when public shareholders want one thing and the nonprofit wants another? If there’s already a confidential filing at the SEC, that question is not theoretical anymore. Okay, step back with me — if OpenAI’s nonprofit is still supposed to be in charge, how does an IPO even work? Who are the executives actually answering to when public shareholders want profits and the nonprofit is supposed to put safety first? It’s a genuinely tricky governance puzzle, and the restructuring OpenAI finished in late October 2025 was supposed to thread that needle — though not everyone thinks it did. The basic setup is this: the for-profit arm became a public benefit corporation, and a renamed OpenAI Foundation nonprofit kept control of that subsidiary while also taking a significant ownership stake. Per TechPolicy Press, that structure also gives OpenAI’s biggest corporate partners — meaning Microsoft — more defined terms on their investment returns. So the nonprofit sits at the top of the pyramid, and in theory the mission still governs. But there’s real tension built in. Researchers looking at OpenAI’s latest IRS disclosure noticed that between 2023 and 2024 the company quietly dropped the word “safely” from its mission statement — it used to say it aimed to build AI that “safely benefits humanity, unconstrained by a need to generate financial return,” and now that word is gone. And per NBC News, the for-profit board is designed to let investors and corporate partners “more easily reap returns” and make a public offering possible. So yes, the nonprofit is in the controlling seat, but the for-profit side is built to be return-friendly — which is exactly the loyalty problem Musk’s lawsuit has been pressing in federal court. So did the trial actually answer the tug-of-war question — nonprofit mission or investors? Not exactly. The jury ruled against Musk on the core theft-of-charity claim, saying he waited too long to sue, and the judge agreed, which clears a major legal cloud over the IPO path. But the governance tension itself doesn’t disappear because there’s a verdict — it just gets kicked back to regulators, the nonprofit board, and eventually public markets. The real thing to watch is how the OpenAI Foundation uses that controlling stake once outside investors are in the picture. That’s where the rubber meets the road. PolisPandit, with John Polonis:
But today’s paywalled piece is an analysis of OpenAI’s IPO and its two masters. How a mission to benefit humanity collides with profitmaxxing owners. How this creates a governance conflict, especially for the foundation at the top that’s supposed to oversee the human part, not the profit one.
PolisPandit — John Polonis, May 27th — has a very specific label for the governance problem we’ve been circling all week: two masters. The nonprofit board keeps formal authority, but once public shareholders show up, you’ve got two legally distinct principals with competing claims on management loyalty, and the confidential S-1 reportedly already filed with the SEC doesn’t name a tiebreaker. Confidential filing. So the SEC is reading this thing right now, and the dual-loyalty problem is no longer a thought experiment — it’s in a registration statement. That’s the answer to the question I asked Monday: if the nonprofit is still nominally in charge, how does an IPO even work? Apparently, the answer is: nobody’s written down who wins yet. And that PolisPandit framing maps straight onto the S-1 disclosure risk. A registration statement has to tell investors which master wins if there’s a conflict. If that mechanism still isn’t named in the filing, that’s not a philosophy problem — that’s a material omission the SEC is going to push on. Polonis also points to the index rule changes — S&P and Nasdaq-100 standards loosening up — which means passive funds can get pulled into buying this structure whether they like the governance or not. You don’t get to opt out because the language is fuzzy. That’s the part that should make people uncomfortable. If you’re following Musk v Altman, you may also like Anthropic Pentagon Watch, a daily briefing on Anthropic’s fight with the DoD over Claude, military AI use, autonomous weapons, and AI procurement blacklisting. Find it wherever you listen to podcasts.
You’ll find links to every story we covered today in the show notes, so if one caught your ear, it’s easy to go deeper when you have a moment.
That’s Musk v Altman Daily for this Thursday. This is a Lantern Podcast.