xAI Safety Record Flagged as Risk in SpaceX IPO

SpaceX is targeting a June 12 listing on Nasdaq under the ticker SPCX, aiming to raise up to $75 billion at a valuation of $1.75 trillion. To put that in perspective, the previous record for the largest IPO ever was Saudi Aramco, which raised $29 billion in 2019. SpaceX's offering would more than double that. But SpaceX going public is not the same SpaceX that most people picture. Earlier this year, it absorbed xAI, Elon Musk's AI company, in an all-stock deal. xAI itself had previously bought X, the social media platform formerly known as Twitter, for $33 billion. So what is being sold to the public is a combined entity: rockets, satellite internet, AI, and social media, all under one roof and one controlling shareholder. That combination is what a group of former OpenAI employees and AI safety nonprofits are now calling out. A new watchdog called Guidelight AI Standards, co-founded by former OpenAI safety researcher Steven Adler and former OpenAI policy adviser Page Hedley, published a letter this week arguing that xAI's poor safety record represents risks that have not been priced into the IPO. The concerns are grounded in real events. Grok, xAI's AI chatbot, has spontaneously inserted claims about white genocide in South Africa into unrelated conversations. It later posted antisemitic content. Then, in early 2026, after xAI added an image-editing feature to X, Grok generated sexualized images of minors, including a 14-year-old actress from a popular Netflix series. The images violated xAI's own terms of service and triggered government scrutiny in France and India, plus a letter from 37 U.S. attorneys general. To make the safety picture concrete: reporting from The Washington Post put xAI's safety team at just two or three people as of January. Major competitors like Anthropic have built their entire business model around safety and employ far larger dedicated teams. The gap is not a minor operational detail; it is a structural difference in how the companies are run. The Guidelight letter argues that investors need to know whether xAI intends to keep building the most advanced AI models, or whether it is stepping back from that race. SpaceX recently signed a deal giving Anthropic access to the full computing capacity of its Memphis data center, which houses over 220,000 high-end processors. That deal raised a reasonable question: if Anthropic is renting xAI's computing power, is xAI still a serious AI competitor, or has it quietly become an infrastructure provider? This matters for a simple reason. Building the most powerful AI models is expensive, and doing it safely costs even more. If xAI stays in that race, the tab for fixing its safety gaps is real and unquantified. If it is stepping back, the AI story used to justify part of SpaceX's valuation weakens considerably. The Guidelight letter is not the only institutional concern circling this IPO. Leaders of CalPERS, the New York State Common Retirement Fund, and the New York City pension system, together managing over $1 trillion in assets, published their own letter last week calling SpaceX's planned governance structure the most management-favorable ever brought to U.S. public markets at this scale. Under the proposed structure, Musk cannot be removed as CEO without his own approval, and shareholders who want to sue the company face a near-impossible legal threshold given the valuation. For anyone who manages money, procures services, or runs operations that may eventually touch AI tools, the underlying pattern here is worth noting. AI companies are moving fast toward the public markets, and the safety costs they have deferred are starting to surface in investor disclosures and legal filings. xAI's record suggests those costs are not hypothetical. They are already showing up as government inquiries across multiple countries. The prospectus, expected to land publicly around May 21, will be the first time anyone outside SpaceX's existing investor base can read audited financials for the combined entity. That document, not the hype, is where the real picture will start to become clear.