Thirteen billion dollars in, twenty-seven percent out, a hundred and thirty-five billion on paper — and one clause that could rewrite all of it the day someone says the letters A-G-I. This is Musk v Altman Daily. Today: what Microsoft's stake actually looks like on paper, a preliminary injunction win that isn't the verdict, and why nonprofit law won't go back to sleep. OpenAI just won in one courtroom while another one's still loading the complaint. We'll get into what that injunction does — and what it very much does not do. Let's start with the numbers, because Value Add VC finally puts them in one place. Microsoft: $13 billion in, roughly a 27% economic interest, now worth about $135 billion. And the 2025 restructuring rewrote those terms. So the S-1 can't just describe the deal — it has to reconcile the pre- and post-restructuring versions in its risk factors. That's a disclosure problem with two sets of math. Right, but the part that keeps me up is the AGI clause. If AGI gets declared, does Microsoft's $135 billion survive, or does the clause basically zero it out? Because you can't file an S-1 that says "our biggest backer's stake evaporates the moment our stated mission succeeds" and expect the SEC to just nod along. Exactly. Investors get told their compute partner could lose its position on a contractual trigger — and now we have the actual trigger language to ask the SEC what disclosure they expect. It's a weird asset. Worth $135 billion until a board declares a milestone, then maybe worth a rounding error. And that folds into a number we've chased all week — Brockman's sworn fifty-billion-dollar compute figure, the Microsoft terms, the foundation's roughly 27% stake. Three valuations that all have to line up under SEC standards, or someone has to explain why they don't. Now, the injunction — because people are conflating it with the jury verdict. Quinn Emanuel got a preliminary injunction for OpenAI. Different standard, different relief, totally different posture. So what did Gonzalez Rogers actually grant? Because a PI is "don't do this thing while we sort it out" — it's not a ruling on the merits. Correct. It doesn't touch the charitable-assets question. That theory survived the verdict, and it's still live in Florida. The injunction is procedurally separate. And that's the trap — one win in one venue, and suddenly it looks like OpenAI's cleaning house. It isn't. Florida hasn't moved an inch. Here's my hangup. The Smith School expert calls nonprofit law "old, sleepy rules." Fine. So why'd OpenAI just sprint to federal court for an emergency injunction? Because "nobody enforces this" and "we needed a judge tonight" can't both be true. The rules are sleepy right up until a charity converts to a for-profit and someone with standing wakes them up. Florida's that someone. And the answer to "can OpenAI go public while this is live" is still no — one venue moved, the one that matters most for the charity fight didn't. So this week moved from "can OpenAI convert" to "what does conversion look like on paper for investors." A $135 billion stake, an AGI clause, and a doctrine everyone called dead, sitting in a Florida courthouse. That's where things stand. Here's Trace Cohen at Value Add VC:
$13B is what Microsoft invested in OpenAI from 2019 to 2023, and after the October 2025 restructuring it holds a ~27% economic stake worth about $135B in OpenAI Group PBC. Microsoft keeps model IP rights through 2032 and a revenue share through 2030, but Azure exclusivity ended and an independent panel — not OpenAI's board — now verifies any AGI claim.
The Value Add VC breakdown finally puts hard numbers where I'd been hedging: Microsoft's roughly $13 billion in from 2019 to 2023, a ~27% economic stake, now worth about $135 billion in OpenAI Group PBC. And the part that should keep an S-1 lawyer up at night — an independent panel, not OpenAI's board, verifies any AGI claim. You've got a $135 billion stake whose terms can flip on a milestone someone else gets to call. Right, and the October 2025 restructuring rewrote the deal: Azure exclusivity is gone, but Microsoft keeps model IP through 2032 and a revenue share through 2030. So the S-1 risk factors have to reconcile a pre- and post-restructuring deal that no longer looks like the one investors think they know. That's the disclosure trap. If AGI gets declared, does Microsoft's stake survive or does the clause gut it? You can't put 'it depends on a panel' in a prospectus and pretend you've handled material risk. From Quinn Emanuel Urquhart & Sullivan:
Quinn Emanuel fought back, rejecting Defendant's claims, exposing Defendants’ deceptive conduct to the US Patent and Trademark Office, and demonstrating that none of Defendants’ various alleged uses were bona fide uses in commerce sufficient to give rise to trademark rights, the victory established that OpenAI’s mark was, as Judge Gonzales Rogers found, “one of the most recognized in the artificial intelligence industry, if not the world.”
So Quinn Emanuel wins OpenAI a preliminary injunction — and I'm bracing for the AGI clause, the Florida AG, the whole capitalization fight. Turns out it's a trademark squabble over the domain open.ai. Right — Judge Gonzalez Rogers enjoined a company and its owner for using 'Open AI' and that domain. It's a Lanham Act case, so nonprofit law doesn't come into this one. And it's a genuinely different legal event from the jury verdict — different standard, different relief. A PI says OpenAI's mark is, in the judge's words, one of the most recognized in artificial intelligence. It doesn't touch the charitable-assets question at all. Which is the part I love. The defendants had the registered trademark, claimed OpenAI sat on it for a year, even claimed OpenAI tried to buy them out. And they still lost — because the judge found none of their uses were real commerce. Quinn Emanuel's argument was sharper than 'we're bigger' — they flagged deceptive conduct in front of the Patent Office. That's what carried it. Clean, narrow legal win. Newswise writes:
Musk told jurors he was “foolish enough” to donate roughly $38 million to what he believed would remain a nonprofit AI research lab—one that could now be valued at more than a trillion dollars.
The Smith School's Samuel Handwerger calls this whole case a fight over 'old, sleepy rules' about who's legally allowed to profit from a charity. Sleepy until thirteen billion dollars wakes them up. Yeah, but here's what bugs me about 'sleepy.' If nobody enforces this stuff, why did OpenAI need to run to federal court for a preliminary injunction at all? You don't sprint to Judge Gonzalez Rogers if the legal system is asleep. Fair — and that's where the doctrine starts to matter. The verdict's behind us, but the charitable-obligation theory survived, and Florida is still the plaintiff standing there, waking up Handwerger's 'sleepy rules.' Right, and stack that against the deal terms from the Value Add piece — twenty-seven percent, a hundred and thirty-five billion. The 'who profits from a charity' question isn't abstract when that's the number sitting on top of a public benefit corp. If you're following the big AI power struggles here, try Anthropic Pentagon Watch — a daily briefing on Anthropic's fight with the DoD over Claude, military AI use, autonomous weapons, and 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 something caught your ear, you can dig into the source material there. That's Musk v Altman Daily for today. This is a Lantern Podcast.