Of the original eleven co-founders who launched xAI with Elon Musk three years ago, only two remain. The deep learning lab is undergoing a significant personnel overhaul as it strives to compete with rivals like Anthropic and OpenAI. Elon Musk insists this rebuilding is intentional, stating that xAI was not built correctly the first time and is now being reconstructed from the foundations up. By most accounts, the process has not been smooth.
Competitive pressure is immediate. This week, xAI co-founders Zihang Dai and Guodong Zhang departed after Musk expressed dissatisfaction that the company’s AI coding tools were not effectively competing with programming assistants like Claude Code from Anthropic or Codex from OpenAI. Musk reported the company held an all-hands meeting focused on catching up, a goal he predicted could be achieved by the middle of this year.
Coding tools are critically important because they represent a key revenue stream. While early user growth for xAI’s Grok was fueled by its lax regulation on generating certain imagery, coding tools are seen as the essential moneymaker for AI labs. This makes xAI’s current lag in this area more than a perception issue; it is a direct business problem.
The personnel changes extend well beyond this week. A month ago, eleven senior engineers, including two co-founders, left xAI following what Musk described as a reorganization for a larger business scale. That effort appears to have been insufficient, as reports indicate executives from SpaceX and Tesla have been brought in to evaluate employees and dismiss those who do not meet the standard. The two remaining co-founders, Manuel Kroiss and Ross Nordeen, along with Musk, face a considerable challenge.
Musk is now casting a wider net for talent. He announced that he and a colleague are reviewing previously rejected employment applications, intending to reach out to promising candidates who should have been interviewed, offering an apology to those overlooked.
For comparison, LinkedIn reports xAI has just over 5,000 employees, compared to more than 7,500 at OpenAI and more than 4,700 at Anthropic.
There is at least one encouraging sign on the hiring front. Andrew Milich and Jason Ginsberg are joining xAI from the AI coding tool company Cursor, where they jointly led product engineering. Unlike Cursor, which depends on other labs for AI models, their move to xAI may signal the importance of direct access to large language models and computing resources, suggesting xAI’s core frontier model remains an attractive asset.
The pressure to show results is both external and internal. Now that xAI is part of SpaceX, and with a potential public offering of SpaceX shares anticipated, the cash-burning unit must demonstrate real user adoption of its Grok model. A struggling AI division is not the narrative Musk wants investors to see.
Longer term, Musk is betting on a vision bigger than coding tools. The “Macrohard” project aims to create an AI agent capable of performing any task a white-collar worker can do on a computer. However, the project’s chosen leader left within weeks, and recent reports indicate Macrohard has been paused.
In response, Musk has drafted another of his companies into the effort, revealing that Macrohard is a joint project with Tesla. Tesla is developing a complementary agent called “Digital Optimus,” a reference to its humanoid robot. In Musk’s description, the xAI language model would direct the Tesla agent as it performs tasks.
This vision is ambitious but not unique. It aligns closely with efforts from other companies, such as Perplexity’s “Everything is Computer” offering for enterprise digital proxies, and similar work on personal agents being pursued at OpenAI.