OpenAI, Tesla mulled buying Cerebras: 4 revelations from latest Musk v OpenAI filing

Extensive email conversations between Musk and Sam Altman and ex-OpenAI leaders Ilya Sutskever and Greg Brockman were filed as part of the lawsuit, first picked up by TechCrunch.

Among the revealing titbits include a debate about potentially changing OpenAI from a non-profit structure, insights into internal dynamics at OpenAI, and concerns about potential conflicts between Tesla's shareholder obligations and OpenAI's mission.

Here’s the best bits:

High on the list of intriguing insights from this latest crop of emails was the suggestion Musk and the OpenAI team proposed acquiring Cerebras, the creator of dinner plate-sized AI chips.

The emails show Illya Sutsekever, who at the time was OpenAI’s chief scientist, proposed OpenAI bought the semiconductor firm, not Tesla, as the automaker “has a duty to shareholders to maximise shareholder return, which is not aligned with OpenAI's mission.”

The emails were dated 2017, the early days of OpenAI, where it was still looking for hardware to power its then-research efforts.

The proposed Cerebras merger never came to fruition, however, with OpenAI instead turning to Nvidia to supply it with hardware to power its AI training efforts after Intel backed out of investing $1 billion for a 15% stake in the then-non-profit.

Cerebras is currently gearing up for an IPO on the Nasdaq, hoping to woo investors looking to back AI hardware firms following the success of Nvidia, with its giant WSE-3 chips some 57 times larger than Nvidia’s H100s.

The decision not to acquire OpenAI compounds Altman’s routine demands for more chips, which has led to reports routinely surfacing that the CEO has pushed the Microsoft-backed AI company to look into developing its own chips.

Altman has also reportedly approached Nvidia’s chip manufacturer TSMC as well as investors in the Middle East and SoftBank to fund his chip project.

Central to the dispute at hand were the concerns around maximising profit versus OpenAI’s (at the time) mission to act as an open counter to Google’s AI efforts.

The latest batch of emails shows that OpenAI leaders were concerned that demands from Tesla's shareholder obligations could impact the then non-profit's mission.

Sutsekever’s email states that the OpenAI team valued having checks and balances, noting that "Sam [Altman] acted as a genuine counterbalance” to Musk.

“The goal of OpenAI is to make the future good and to avoid an AGI

Dictatorship,” the emails read. “You are concerned that Demis [Hassabis, CEO of Google-owned DeepMind] could create an AGI dictatorship. So do we. So it is a bad idea to create a structure where you could become a dictator if you chose to, especially given that we can create some other structure that avoids this possibility.”

The emails reveal that Musk wanted to be CEO of OpenAI, not an uncommon demand from a man who holds multiple leadership positions at myriad companies. At the time of writing, Musk is CEO and chief engineer of SpaceX while also CEO of Tesla whilst simultaneously CTO and executive chair of X, and founder of several other entities.

While Musk’s CEO designation differs from how a traditional executive would operate in that particular role, his apparent demand to be CEO is not common with his prior pattern of wanting a top title at firms he invests in, like listing himself as a co-founder of OpenAI.

Sutsekever’s email suggests that Musk said he “needed to be CEO of [OpenAI] so that everyone will know that you are the one who is in charge, even though you also stated that you hate being CEO and would much rather not be CEO.”

The email from Sutsekever expressed serious concerns about Elon Musk potentially gaining “unilateral absolute control over the AGI.”

“You stated that you don't want to control the final AGI, but during this negotiation, you've shown to us that absolute control is extremely important to you,” the emails read.

The emails provide a stark view of the clear end of OpenAI’s relationship with Musk.

Musk left OpenAI in 2018 after disagreeing with its pivoting away from being a non-profit.

However, in an email dated September 2017, Musk told the OpenAI team: “I've had enough. This is the final straw.”

“Either go do something on your own or continue with OpenAI as a non-profit,” Musk wrote. “I will no longer fund OpenAI until you have made a firm commitment to stay or I'm just being a fool who is essentially providing free funding for you to create a startup. Discussions are over.”

Musk sued OpenAI and its co-founders back in February, claiming the ChatGPT makers broke a founding promise to operate as a non-profit in a breach of contract after it partnered with Microsoft.

Musk was among the co-founders of OpenAI, providing finances to the then-non-profit to support its ultimate goal of developing AGI.

The Tesla, X, and now xAI CEO argued in his rather peculiar lawsuit that the dispute focused on benefiting humanity and that by OpenAI moving away from its non-profit pledge, it risked artificial general intelligence or AGI becoming in the hands of corporations.

Musk then dropped the lawsuit in June days before OpenAI were due to request the case to be dismissed.

OpenAI had published a series of emails shortly after Musk filed suit that painted him as the one wanting the then-nonprofit to switch to for-profit, even suggesting merging OpenAI into Tesla.

Musk then reignited the dispute in August by filing a new lawsuit, double the length of the original, alleging fraud, racketeering, false advertising and unfair competition in addition to the original breach of contract claims.

Musk has routinely attacked OpenAI throughout 2024, notably threatening to ban staff from his companies from using Apple devices following OpenAI’s partnership with the iPhone maker, claiming the devices posed “an unacceptable security violation.”

After departing OpenAI, Musk went on to found his own AI startup, xAI, which is trying to build Grok, a rival to ChatGPT that’s more risqué with its responses.

He’s spent the best part of 2024 snapping up GPUs to build what he hopes will be the most powerful supercomputer in the world in Memphis, Tennessee to power Grok training. xAI recently announced plans to double the size of its already 100,000 GPU cluster as it gears up to train Grok 3, which Musk said is expected to launch by the end of the year.

Musk’s plans for xAI’s supercomputer to be the most powerful would have to compete with El Capitan, the new HPE and AMD-based system which took that crown from Frontier in the latest Top500 list.

Elon Musk to lead Trump's cost-cutting DOGE 'department'

Musk to double xAI’s Colossus cluster to 200K GPUs for Grok 3 training

OpenAI vs xAI: The battle for AI supremacy