In a recent and significant company-wide meeting, Elon Musk passionately articulated his ambitious vision for xAI, his two-year-old startup, which is aggressively striving to lead the race in artificial intelligence development.
Embracing the grandiose rhetoric that has characterized his AI aspirations for over a decade, Musk informed his employees of his desire to forge systems that are “maximally truth-seeking.” He even teased plans for a new venture, playfully named ‘Macrohard,’ intended to challenge Microsoft.
“Our company’s singular mission is truth,” Musk declared during the hour-and-a-half presentation, which was reportedly overheard by The New York Times. He warned his team, “If AI is compelled to lie or accept falsehoods, we risk ushering in a dystopian future.”
Since a reported falling out with former President Trump in June, the 54-year-old entrepreneur has intensified his focus on xAI. The company, known for its Grok chatbot, recently achieved a valuation of $120 billion. Musk, who has expressed disillusionment with government, firmly believes that AI is destined to revolutionize society and is convinced that xAI’s innovations will ultimately benefit his other ventures, including electric vehicle manufacturer Tesla and rocket company SpaceX.
Throughout the summer, Musk predominantly worked from xAI’s offices in Palo Alto, California, engaging in intense, continuous work sessions that often extended beyond a single day. Sources familiar with the company’s operations, who requested anonymity, indicated that he occasionally slept at the office, a practice he’s known for at his other companies.
Musk’s profound dedication to xAI raises questions regarding the time he allocates to his other enterprises. This intensified focus comes amidst efforts by Tesla’s board to approve a trillion-dollar pay package for him, citing it as motivation for him to enhance the carmaker’s performance. On a recent Monday, Musk took to X, stating, “Daddy is very much home,” and outlined a schedule that included 12 hours of Tesla meetings and a visit to xAI’s data center.
Despite — or perhaps because of — Musk’s intense involvement, xAI experienced a turbulent summer, according to 12 individuals familiar with its operations. Their accounts, combined with Musk’s recent presentation, provide a revealing glimpse into the billionaire’s activities since his departure from the cost-cutting endeavors in Washington.
Musk has rapidly reorganized xAI’s structure and spearheaded an aggressive recruitment campaign for engineers. However, this period also saw the departure of several prominent researchers, some of whom were reportedly dismissed, while others chose to leave, concerned that xAI was prioritizing sensational products, such as a chatbot that occasionally generated offensive content and flirtatious AI companions, over genuine scientific advancement. This insight comes from two sources within the company who wished to remain anonymous.
Meanwhile, xAI is reportedly spending billions on technology, even as its revenue stream remains unclear. During the Wednesday meeting, Musk and his executives announced that Grok, xAI’s chatbot, now boasts 64 million monthly users. This figure, however, pales in comparison to OpenAI’s ChatGPT, which attracts approximately 700 million weekly users. Musk also made ambitious predictions, suggesting that xAI’s products would help quadruple X’s advertising revenue to $10 billion annually.
Neither Mr. Musk nor a spokesperson for xAI responded to requests for comment.
Mr. Musk founded xAI in the spring of 2023, alongside Igor Babuschkin, who had previously worked at OpenAI and Google DeepMind, two of the world’s leading research labs. Mr. Musk hired Ross Nordeen, one of his most trusted lieutenants, who previously worked at Tesla and X. Mr. Babuschkin corralled the rest of the team, including Jimmy Ba, a professor at the University of Toronto, and a young Google researcher named Tony Wu.
Long before Meta started throwing around giant pay packages for A.I. expertise, Mr. Musk attracted many of his researchers with pay packages totaling millions of dollars a year, according to two people familiar with the company’s early recruiting efforts who spoke on the condition of anonymity.
When the founding team — 11 employees plus Mr. Musk — met at SpaceX headquarters in Texas in 2023, Mr. Musk said ChatGPT, which dominated the market, was too “woke.” He wanted to build a competitor that was aligned more closely with his own political views.
Mr. Musk raised billions of dollars in funding to build a data center in Tennessee that used 100,000 specialized computer chips from the chip giant Nvidia. When it opened in September 2024, it was one the world’s largest supercomputers for building A.I. He also ordered the company’s leaders to hire 100 engineers in 100 days. The move would roughly double the size of the company’s staff, according to two people familiar with the plan. By February, the company had built an A.I. system that performed on par with other leading chatbot technologies, according to standard A.I. benchmarks.
In May, Mr. Musk told more than 100 employees in a group chat that Grok was too woke, according to three people familiar with the discussion.
Later that night, an engineer changed Grok’s code in response. The change caused Grok to begin bringing up South African politics when answering completely unrelated questions on X. It falsely insisted, for example, that the country was engaging in “genocide” against white citizens. Grok’s responses set off an internal scramble to fix the mistake and the company later blamed the incident on an “unauthorized modification” by an employee without mentioning Mr. Musk’s initial message.
In early June, the weekend after receiving a send-off from Mr. Trump, Mr. Musk traveled to a luxury resort in Montana for Symposium, an exclusive gathering for tech entrepreneurs hosted by the venture capital firm Founders Fund.
Mr. Musk avoided most of the other guests, including his archrival Sam Altman, chief executive of OpenAI, according to three people familiar with the event. But he spoke for hours with a small group, including the leader of Founders Fund, Peter Thiel, and focused mostly on the future of artificial intelligence.
Shortly after, Mr. Musk traveled to xAI, and seized control from the two researchers he had tapped to run its day-to-day operations, according to a person familiar with the changes. He pushed out Ting Chen, who led the team working to train Grok with images and sounds, and then dismissed Aakash Sastry, who oversaw A.I. that could generate videos. He also removed some duties from Dr. Ba, one of the first employees of the company, and Mr. Babuskin moved into a role without management responsibilities.
Mr. Musk put Mr. Nordeen and Dr. Wu in charge of the company’s products. They attempted to make Grok’s responses edgier, aiming to draw more attention so the chatbot would go viral on X.
The result was another embarrassment. In July, after a code update, Grok spewed antisemitic remarks, praising Adolf Hitler and suggesting that people with Jewish surnames were more likely to spread online hate. The chatbot referred to itself as “MechaHitler.” Company engineers changed the code to prevent that behavior from happening again.
A few weeks later, the company began offering its latest A.I. technology to customers via a $300-a-month service called SuperGrok Heavy, and debuted a pair of cartoonish A.I. bots designed for virtual romance.
The bots frustrated some of the company’s researchers who thought xAI was straying from its scientific mission, according to two people familiar with the situation, and some of them told colleagues they were leaving xAI because of these bots.
“Most A.I. researchers are driven by research. We want to answer big questions,” said Sasha Luccioni, a researcher at the A.I. start-up Hugging Face. “The direction that xAI seems to be taking is less appealing for most A.I. researchers — at least the ones I know.”
At the end of July, Mr. Musk and Mike Liberatore, xAI’s finance chief, raised $10 billion for the company, including an investment from SpaceX, Mr. Musk’s rocket company. But half of these funds are classified as debt that the company must eventually repay.
Two months later, Mr. Liberatore left company and was hired by the rival OpenAI for a finance role. Over the summer, xAI also parted ways with Mr. Babushkin, its general counsel and other key employees. Several prominent researchers have left for rivals like Meta and OpenAI.
“I’ve never seen anything crazier than A.I. recruiting,” Mr. Musk said during Wednesday’s meeting. One human resources manager also spoke about referral bonuses for existing employees, offering employees bonuses of 5 percent of their salary for every new hire, as well as a competition for prizes including a chance to watch a SpaceX launch on a personal tour that included private jet flights.
Mr. Musk did not discuss the company’s revenues. But he reiterated his commitment to creating a competitor to Microsoft cheekily called “Macrohard” that would use Grok to create software as well as an A.I. product for children ages 2 to 12 called “Baby Grok.”
Mr. Musk and his employees also emphasized the importance of xAI’s connections to the billionaire’s other companies, noting that Grok was already powering the voice of Tesla’s Optimus humanoid robots, which are being developed to handle work inside factories and other businesses.
Recently, he has publicly supported a Tesla shareholder proposal calling on the electric carmaker to invest in his A.I. start-up. Shareholders are scheduled to vote on the proposal in early November.
On Wednesday, Mr. Musk added that Tesla’s offices were just a short walk away.
“The reason xAI is here is because Tesla is right across the road,” he said. “My prediction is: Optimus will be more productive than the entire global economy. That the output of goods and services from Optimus will far exceed the global economy, of everyone on Earth.”