Elon Musk

People started quitting OpenAI. Altman went to Brockman’s house and wanted to figure out a way to stabilize OpenAI. He had calls with board members about coming back to OpenAI. “Although I was still very angry” and thought he’d have to “clean up a mess I didn’t make,” he said he was willing to come back under other conditions.

All the board would tell him was that he wasn’t consistently candid and they weren’t going to get into why. Altman was completely shocked, he says. He also told them that their plan to announce it via a blog post would throw things into chaos. “If this is the decision, this is a terrible way to execute it,” Altman said he told them. The board told him it was too late to stop it.

2023 was “the beginning of the inflection.” ChatGPT had been introduced, and “it became clear to us we would need a lot more compute.” They needed it for both research and for the models being used by the public. Around then, Shivon Zilis resigned from the OpenAI board. Shortly after that, Musk announced xAI, and Altman says there were”a lot of efforts to recruit our employees” and “negative tactics from Mr. Musk toward us.” Musk’s lawyers don’t like this but over their objections, Altman “started to hear rumblings” about litigation.

...and it’s a little catty. (I live!) At one point, Altman says that Zilis told him Musk had “front-runner-itis” — but there’s an objection that stops Altman from telling the rest of this story. Altman looks slightly disappointed. We then hear that Zilis advised Altman on how to engage Musk so that Musk wouldn’t “bash us on Twitter.”

Musk didn’t raise any objections, though. And then he sent the infamous message where he rated OpenAI’s chances as zero. Altman appears to be concentrating hard on his testimony but is coming across as being a little bewildered about why he is here at all — but maybe that’s just how his eyebrows look at all times.

We’re getting testimony about emails and meetings Altman had with Musk to try to walk him through the for-profit. They reviewed documents together at the meeting and then emailed him the term sheet that Musk testified he didn’t read.

It was when Altman met with Musk and Zilis to discuss plans for for-profit meetings. Zilis texted after the meeting to say she was glad they had the meeting to let Musk think about “the investment thing so it won’t irk him later.”
A good vibes meeting means a long conversation of Musk”showing us memes on his phone.”

He learned in 2022 that Musk was the father of her kids, and keeping her on the board was “a close call for me personally because she had sort of told us that Mr Musk was playing a more involved role than originally intended and that they were spending more time together.” On the other hand, Altman says he thinks highly of Zilis “and valued her counsel.”

Musk’s departure from the board had a mixed result on morale. “Mr. Musk is a well known figure and known to be fairly mercurial and people wondered if he was gonna try to take a vengeance out on us or something.”
On the other hand, people were relieved to be rid of him. “I don’t think Mr Musk understood how to run a good research lab.,” Altman said, “He had demotivated some of our most key researchers.”

He “didn’t want to be associated with something he couldn’t control and didn’t think would succeed.” Additionally, Musk wanted to work on AI at Tesla and didn’t want to be conflicted.

“We were kind of running the org on a shoestring” and had “an extremely short runway of cash,” Altman said. OpenAI didn’t meet its fundraising goal of $100 million in 2018, raising only a hair under $50 million. Major donors are Aphorism Foundation (Reid Hoffman), Fidelity Charitable, Gabe Newell, Good Ventures Foundation (Dustin Moskowitz), Amazon Web Services and, hilariously, Alameda Research (FTX / Sam Bankman-Fried).

“The only path - the best path that he saw - was for OpenAI to become part of Tesla,” Altman says. We are now looking at text messages. Musk “remains very open to you joining the Tesla board as part of this,” Musk’s subordinate Sam Teller wrote. And, also, Teller said, “regardless of how these conversations about OpenAI shake out, he is committed to building a stronger AI team within Tesla.” Altman said, “I viewed a vague, like a lightweight threat in there” that Tesla would do it with or without OpenAI.

Altman asked what would happen if Musk died. Musk said, “I haven’t thought about it a ton, but maybe control should pass to my children,” Musk replied. We also see an email where Altman says, “I desperately want to see this work with Elon... but I am worried about control. I don’t think any one person should have control of the world’s first AGI.” He says he’d be open to creative structures - like Musk having control up to a certain milestone.

Someone — Altman doesn’t remember whether it was Musk or his subordinate, Sam Teller, said it — told Altman that Musk had “long since decided” he would only work on companies that he controlled. “Mr. Musk felt very strongly that if we were going to form a for-profit he ended to have total control over it initially and this was because he only trusted himself to make non-obvious decisions that were going to turn out to be correct,” Altman says.

He is being sworn in now.

Having the ability to sell their equity let them finance their activities, such as AI research into Alzheimer’s research.

He is now testifying about Musk’s attempt to buy OpenAI using xAI, which happened after this lawsuit was filed. “I was surprised” by it, Taylor says. “Ostensibly this lawsuit is about our nonprofit purpose and mission adn this proposal was to acquire this nonprofit by a group of for-profit investors, which felt contradictory to the spirit of this lawsuit.” The OpenAI Foundation board rejected the bid because they didn’t feel it was appropriate for one person to control the mission

But also about some light character assassination. It looks like Musk now has a broader strategy to try to make Altman a liability to OpenAI.

He has not managed to do this at all. He is speaking rapidly, in a monotone.

“We’re decidedly not cash-flow-positive today.” The company has not generated any profits to date.

Because LLMs keep stealing people’s work, lol. Anyway he was talking about the OpenAI deal with Reddit, which was done to avoid litigation.

He is in a gray suit and gray tie. I am expecting more pleated khaki pants testimony. He was also the chair of Twitter’s board when it was acquired by Musk.

Also, Musk gave no guarantee his proposed control of the board would diminish over time. “I found it to be aggressive because I knew that Mr Musk had many other obligations in many other companies that the was running that were much larger than OpenAI,” he said. He also didn’t like the proposal that Tesla take over OpenAI. “It would be on some level, it would be like, it would kill a dream,” he said. “When one starts a company, one has dreams for a company to flourish and do different things, and in general being absorbed into another company means to give up that dream.”

We are now at “the blip” again. The board chose “not consistently candid” carefully, he says. Sutskever said he prepared a document of incidents with Altman, with some other people at OpenAI. Altman has a pattern of lying and pitting executives against each other, “this leads to tremendous loss of productivity,” trust, and difficulty creating safe AGI.

Our next witness is Ilya Sutskever. This promises to be more interesting, I think.

He says he provides suggestions a lot. And they’re just suggestions, including in the case of OpenAI. While Microsoft put forward 14 names, none of those names were added to the new OpenAI board except Sue Desmond-Hellman. CEO of the Bill and Melinda Gates Foundation, who was added later.

He’s explaining that firing a CEO is “a fairly big things” and when he didn’t get any details on why Altman was fired, he felt that OpenAI’s board was “sort of amateur city as far as I was concerned.”

We also discover that Nadella has no idea if Altman and Musk talked about Musk’s mean Microsoft/OpenAI tweets. As a side note, we are referring exclusively to “Twitter” and “tweets” even for posts after the X rebrand.

I recognize Musk has identified this as the spot where he was convinced he’d been swindled. But the main thing I’m hearing here is that OpenAI was unstable. I don’t know what the correct way to respond to that kind of chaos is, especially if you have a partnership with the company that appears to be rapidly imploding. Trying to suggest that Nadella’s attempts to stabilize the company and preserve his investment was somehow improper seems weird? This is a genuinely odd situation, not necessarily one you run into in business school. (Well, until then anyway.)

“Pretty definitive statement as to why they fired him, right?” Well, Mr. Molo, no. I remember a great deal of speculation about what this meant; for instance, if Altman had gotten up to something criminal. One of those moments where I wonder what the jury is thinking, since I doubt they recall any of this.

Nadella is cooler than everyone else on cross, but he’s still getting worked up. Molo just got told by the judge that his question was argumentative. Molo is yelling about how the risk MSFT took on OpenAI was “prudent.” Nadella notes it’s still a risk. “At the time the was a risk it could to to zero,” Nadella says. “It was a calculated risk.” Molo, of course, is focused on projected return, of $92+ billion.

Molo is asking about this line Nadella wrote in a 2022 email, and it’s very funny. Nadella has explained a couple times that to him it’s about IP rights. “The context for me was making sure that Microsoft was benefitting from the IP rights that we had because that’s what happened in the case of Microsoft and IBM.” Molo then says, didn’t Microsoft become more prominent and important than IBM? Nadella agrees.

Nadella’s attitude is not notably different than it was on direct. We are talking about donating commute to OpenAI from Azure services. From a 1/12/18 email: “The 2016 deal, where they agreed to pay $10M for $60M in Azure services was projected at a $15M loss over 3 years, given assumed usage profile. They’ll have consumed all the usage in ½ that time.” The funny recurring theme here through the trial is the gaping maw of compute required is always bigger than whatever huge number people had projected. Interesting to know ahead of AI IPOs!

He misstated at first that he didn’t currently serve on any such boards, though he is a trustee at the University of Chicago, which is a nonprofit.

Nadella is being asked about possibly the most baller thing he’s ever said. “Below them” means compute. “Around them” means API. “Above them” means products like Copilot. “The question was being asked, what happens if OpenAI disappears, will everything crumble? So I was trying to reassure everyone,” he says. The problem was that until he said this to Kara Swisher, “the drama of what was going on was drowning out what the customers care about.”

He was being asked if at any point Musk contacted him to say that the OpenAI deals with Microsoft violated any agreement Musk had with OpenAI’s nonprofit. Musk did not. “Does he know how to contact you?” his lawyer, asked. “He does.”

The partnership itself was “a fairly big decision for us” because “to make the call we’d use some of the scare resources we had” on it. But Microsoft accepted the risk of investing in Open AI because Microsoft has a “core ethos as a platform and partner company,” he says. “So if you find partners you can create these win/win with, it’s great to make them longterm stable.”

He looks very nice. I am fully expecting his testimony to be the equivalent of a pair of pleated khaki pants.