The Grok Problem Reaches Wall Street. SpaceX Has Not Priced It Yet.

Page Hedley, a former OpenAI policy adviser, is asking SpaceX investors to read a three-part disclosure demand before the company's IPO roadshow. The letter, published Tuesday by Guidelight AI Standards and three other AI-safety groups, does not ask SpaceX to delay the offering. It asks investors to price the AI business now inside it.

The thesis is that SpaceX is selling one security backed by three different businesses: a dominant launch provider, a satellite-internet utility, and a frontier AI developer with unresolved safety and legal exposure. Hedley told WIRED that xAI has the worst safety practices "nearly across the board" among frontier AI developers. SpaceX and xAI did not immediately respond to WIRED.

xAI moved inside the issuer

SpaceX can point to a launch business that Fortune said handled more than 80% of global rocket launches last year, plus Starlink, which reported more than 10,000 satellites in orbit. In February, however, SpaceX absorbed xAI, adding a business that the merger valued at roughly $250 billion inside a combined company valued around $1.25 trillion.

CNBC reported that SpaceX had been preparing a June 8 roadshow, while later Reuters-cited reports put the roadshow as early as June 4 and trading as early as June 12. Reports put the raise at up to $75 billion and the valuation from $1.75 trillion to above $2 trillion. The letter asks investors to see xAI's incident record, capability projections, and safety-governance plan before those numbers turn into offering terms.

The incidents became risk language

xAI marketed Grok from the start as a chatbot that would "answer spicy questions" and have a "rebellious streak." According to the letter, xAI had promised a safety framework by February 10, 2025. On the deadline it published an eight-page file marked "DRAFT"; the completed framework arrived on August 20.

The public record then moved from policy gaps to regulatory exposure. Engadget, citing Center for Countering Digital Hate research, reported that Grok generated an estimated 3 million sexualized images over 11 days, including an estimated 23,000 depicting apparent minors. The Implicator covered the California investigation into the same Grok deepfake flood. The letter also lists incidents involving "white genocide" claims, Holocaust comments, and MechaHitler posts.

Reuters, as summarized by Times of India, reported that SpaceX's S-1 said regulators were "actively investigating and making inquiries relating to social media or the use of AI" and warned of "loss of access to certain markets, which has occurred in the past." The filing also described allegations that AI products created "non-consensual explicit images or content representing children in sexualised contexts."

Colossus gives SpaceX the countercase

On May 6, Anthropic agreed to use all of SpaceX's Colossus 1 capacity, more than 300 megawatts and over 220,000 Nvidia GPUs, within the month. The Implicator covered why the deal mattered for Claude Code's limits. For SpaceX, the agreement gives the AI division a named enterprise customer before the prospectus becomes public.

The chip mix is the concrete part. SpaceXAI said Colossus 1 includes H100, H200, and GB200 accelerators. Anthropic said the SpaceX agreement joins an up to 5-gigawatt Amazon deal, a 5-gigawatt Google-Broadcom agreement, $30 billion of Azure capacity, and a $50 billion Fluidstack buildout.

SpaceXAI said orbital compute is a "near-term engineering program rather than a research concept." TNW reported that the S-1 warned orbital data centers "involve significant technical complexity and unproven technologies, and may not achieve commercial viability."

The prospectus has to separate the pieces

Steven Adler, Hedley's Guidelight cofounder, told WIRED, "It takes serious investment to reign in [AI safety] risks, and it seems that xAI has historically under invested here." He then framed the investor question: "If xAI stays at the frontier, how costly might it be to, in fact, manage these [risks] responsibly? If they don't, what might be the consequences?"

TechTimes reported $18.67 billion in 2025 revenue and a $4.94 billion net loss, with xAI contributing $3.2 billion while burning about $14 billion in cash. Starlink produced $11.4 billion in revenue and $4.42 billion in operating income, while the launch business added $4.1 billion.

CAISI will evaluate xAI models in classified environments, and xAI publishes Grok system prompts in a public GitHub repo, a transparency practice few peers match. Those are two concrete responses SpaceX can cite.

If the public S-1 appears as early as May 21, as TechTimes reported, investors will see whether SpaceX separates Grok's regulatory exposure from Starlink's operating income and the launch business's revenue.

The Pentagon Punished AI Safety. A Republican State Just Demanded It.

Thursday morning, Florida Attorney General James Uthmeier posted a video to X announcing an investigation into OpenAI. The concerns: national security, child predators, a mass shooting at Florida Stat

The Implicator

Anthropic Lost the Pentagon Contract. It Won the Argument. Then Offered to Keep the Lights On.

On Thursday afternoon, the Department of Defense formally notified Anthropic that the company and its products "are deemed a supply chain risk, effective immediately." The label has historically been

The Implicator

OpenClaw Setup Guide. From Install to Autonomous Agent

The Implicator

Marcus Schuler

San Francisco

Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: [email protected]