SpaceXAI is running into some serious trouble with its staff. Since February, more than 50 top researchers and engineers have walked out the door. This isn’t just a few junior developers; we are talking about key leaders who were building the brains behind the Grok AI. This massive drain of talent comes at a critical time when the company is trying to prove it can compete with the biggest names in the tech world.
Competition for AI experts is fierce right now, and rivals are basically waiting at the exit with open wallets. Mark Zuckerberg’s Meta has already hired at least 11 former SpaceXAI employees. Another 7 workers left to join Thinking Machine Labs, the new startup created by former OpenAI boss Mira Murati. Losing this many people in just a few months is a massive blow to any company, especially one trying to win a technology war that requires thousands of specialized minds.
This exodus started right around the time SpaceX acquired xAI. Elon Musk owned both companies, but he decided to merge them into one giant entity back in February. He officially changed the name of the combined group to SpaceXAI earlier this month. While Musk likely hoped the merger would make things more efficient, it seems to have caused a lot of friction. He brought in a new leadership team, and many of the old guard simply did not like the new direction or the change in management style.
The most worrying part for the company is the state of the “pre-training” team. In the world of artificial intelligence, pre-training is the most important first step. It is where the computer learns the basics of human language and logic. After the team lead, Juntang Zhuang, quit, most of the other researchers followed him. Now, only a tiny handful of people are left to do the work that used to require dozens of experts. Without a strong pre-training team, it is incredibly hard to build a model that can actually challenge ChatGPT or Google Gemini.
Inside the office, the pressure is supposedly sky-high. Sources familiar with the situation say Musk set deadlines that were simply impossible to meet. Because he wanted results so fast, the team had to cut corners on the latest version of Grok. If the accuracy or the “logic” of the AI drops by even 1.5% compared to its rivals, it could be a disaster for a company that has already spent over $10 billion on expensive computer chips. Engineers generally do not like being forced to release “half-baked” products just to meet a date on a calendar.
Musk is famous for his “extreme work” culture, which he has implemented at Tesla and X. He expects people to work long nights and weekends to achieve his goals. While that hardcore style helped him build rockets and electric cars, it is proving more difficult in the AI world. Researchers in this field are in such high demand that they can get $1 million salary offers from almost any other company. They do not have to put up with stressful conditions if they do not want to, especially when other firms offer better work-life balance.
Money is also playing a big role in why people are leaving now. SpaceX regularly lets employees sell their private shares to other investors through tender offers. This means many of these engineers are already sitting on millions of dollars in vested stock. With talk of a massive IPO worth over $100 billion in the future, many workers feel they have reached the “light at the end of the tunnel.” Once they know their bank accounts are safe, they are much more likely to leave a high-pressure job to find something more relaxed.
Meta is being particularly aggressive in its hiring strategy. They are looking to expand their own AI capabilities and are happy to take anyone who has experience working under Musk’s intense systems. Even though SpaceXAI claims it is still building the most powerful AI in the world, the loss of over 50 experts makes that claim look shaky. You cannot build world-class technology with just a few people and a lot of hardware; you need the human minds to make the silicon smart.
The departures include leaders from very specific and vital departments, like Grok’s voice features and “world models.” World models are what help an AI understand how the physical world works, which is vital if Musk ever wants his robots to walk around safely. If these leaders are gone, it delays the entire roadmap for both SpaceX and Tesla. It shows that even a billionaire like Musk cannot keep talent if the environment becomes too frustrating or the goals become unrealistic.
As we move into the middle of 2026, SpaceXAI has to find a way to stop the bleeding. If they keep losing 10 or 15 people every month, they will eventually run out of the experts they need to maintain their massive servers. Musk might be able to buy all the computer chips in the world, but if Meta and Thinking Machine Labs take all his best people, SpaceXAI will struggle to stay relevant in a market that moves at the speed of light.
