The OpenAI Exodus
This makes no sense. Or does it?
Back in September 2024, when Americans still had a glimmer of hope for the future, OpenAI CTO Mira Murati randomly quit to “create the time and space to do [her] own exploration.” If you don’t remember Murati, she has some of the most hilarious interviews in the AI world, where she fundamentally misunderstands art, the modern workplace, how AI works, and where OpenAI got its stolen data from. She is the epitome of the out-of-touch tech bro. Then, in October, it was leaked that she was in talks to raise $100 million to create a stealth startup to challenge OpenAI. Well, she has just broken cover and announced her new startup, Thinking Machines Lab, which has poached dozens of leading researchers from OpenAI. This might seem like business as usual in the cutthroat world of corporate America, but if you just dig a little deeper, it makes no sense at all! Why abandon and then challenge a company that seems to be gunning for the top and aggressively destroying all of its competition?
Let’s start with Thinking Machines Lab. Mira Murati unveiled the startup with $100 billion in backing last Tuesday, with her as the CEO, OpenAI’s co-founder and leading AI researcher John Schulman as chief scientist, and OpenAI’s ex-chief research officer Barret Zoph as CTO. These three were crucial to propelling OpenAI to the top of the AI sphere, and all only left OpenAI in the latter half of 2024. But it isn’t just them who are jumping the Altman ship. Thinking Machines Lab has poached 20 leading researchers from OpenAI and 10 from Meta and Mistral.
This represents a major exodus from OpenAI.
So, what makes Thinking Machine Lab different from OpenAI? Well, they intend to “make AI work for [people’s] unique needs and goals” and to create AI systems that are “more widely understood, customisable, and generally capable” than those of OpenAI. They will do this by focusing on making “multimodal” systems that “work with people collaboratively.” This, to me, sounds like marketing bumph for being open source, allowing people to see and customise their AI’s code, like DeepSeek (read more here).
As such, it doesn’t seem like Thinking Machine Lab has any technological advantage or a unique approach to building AI, as its goals are painfully close to those of OpenAI.
So, why have so many major figures jumped ship?
Thinking Machine Lab has exponentially less backing than OpenAI and doesn’t seem to be unique in the AI world at all.
Meanwhile, OpenAI is effectively receiving the largest private investment in history with the $500 billion Stargate Project and has a virtual monopoly on the generative AI market. Altman is even stoking the speculation fires by claiming they are getting close to creating AI superintelligence. This means that OpenAI’s valuation is set to skyrocket over the coming years. With the company’s transition to a for-profit model, employees, particularly researchers and higher-ups, will be getting shares!
It makes no sense for these people to leave OpenAI. I mean, Project Stargate and the move to a for-profit model and giving employees equity started months before Murati left.
Make no mistake, these guys are leaving millions of dollars on the table. We have to ask, why?
I can see two reasons.
Firstly, OpenAI’s mega valuation of over $157 billion is not based on business fundamentals. Thanks to DeepSeek, exponentially rising AI training costs, and the diminishing returns of AI training, OpenAI currently has zero viable paths to profitability.
Project Stargate is also transparent about tax write-offs for its investors that will boost their other investments (read more here). It also seems highly likely that the AI OpenAI can build with this infrastructure won’t be that much better than their current ones. As such, Stargate might only function as hype and dramatically increase its already staggering operational expenses.
OpenAI is also gunning for hype, with Altman pushing the notion that they are close to developing superhuman AI intelligence, or AGI, rather than actually fixing the problems with current AI models. This is half the reason DeepSeek could dunk on them with a fraction of the investment.
OpenAI is set to run at a loss of tens of billions of dollars per year over the coming years. But Altman doesn’t care; he seems to be focused on maximising the valuation, doubling down on hype and speculation.
However, by pushing demonstrably false AGI claims and taking such dramatically large investments, OpenAI is opening itself up to the impact of reality. It makes OpenAI one giant high-risk bet on the continuing separation of our economy and stock market from reality.
So, I suspect these former employees have jumped ship because this bet is too risky for them. Machine Learning Labs’ smaller budget and lower operational costs while still associating with the people who created OpenAI is what makes the company so promising. Even if they don’t have any technological lead, they can ride the speculation and hype but without opening themselves up to the reality and risks of obviously false claims, moronically giant investments, and multi-billion dollar annual losses. It’s a smaller bet, but it is a far safer one.
The second reason is leadership. Murati, Schulman, and Zoph were all heavily involved in Altman’s temporary ousting as OpenAI CEO in 2023. Murati, in particular, had voiced concern about Altman’s toxic and risk-happy leadership. Now, they may have disagreed with Altman’s high-risk approach to OpenAI, as we have just discussed. But it seems they also don’t like his inability to work collaboratively or take feedback, creating a very billionaire “I know best” dogmatic leadership, which is not conducive to growth at all. Naturally, they might be adding two and two together and see OpenAI as a sinking ship.
Either way, this exodus is a major problem for OpenAI. It’s also yet more evidence that the tech world isn’t operating in reality and instead is desperately trying to exist within a world of pure speculative cyberspace.
Like many, I am so exhausted by this AI stupidity. Why can’t these guys just touch grass, calm down, and actually find a way to make reality a better place?
Thanks for reading! Content like this doesn’t happen without your support. So, if you want to see more like this, don’t forget to Subscribe and help get the word out by hitting the share button below.
Sources: Tech Crunch, FT, Reuters, Fortune, Reuters, Upstock, Fortune, Will Lockett, Will Lockett, Will Lockett, Will Lockett