The first entry in this story was filed on June 5, when SpaceX, OpenAI, and Anthropic were all still private companies lining up for public markets. Since then, one of them has gone through.
SpaceX began trading on June 12 under the ticker SPCX. It raised $75 billion, the largest single public offering in history. The stock closed its first day up 19% at $161, briefly hit $225 by June 16, and trades near $185 today, giving the company a market value of around $2.4 trillion. To give that a frame: it now sits just below Amazon. Elon Musk, whose shares carry 10 votes each versus one for publicly sold shares, became the world's first trillionaire on day one.
The financial picture inside SpaceX is not what the headline valuation might imply. Starlink, the satellite internet business, generated $4.4 billion in operating profit in the first quarter of 2026 alone, at a 63% margin. That is a genuinely excellent business. It is also the only profitable part of the company. The xAI division, which Musk folded into SpaceX earlier this year, posted a $6.35 billion operating loss in all of 2025, followed by another $2.47 billion loss in the first quarter of 2026. Starlink's profits are currently covering those losses in full. Buyers of SPCX stock are effectively buying one strong business and one expensive AI bet, bundled together with no ability to separate them and no meaningful vote on how either is run.
Independent analysts have been direct about the valuation. Research firm Morningstar values the stock at $63 per share; it closed its first day at $161. CFRA initiated coverage with a sell rating and a 12-month target of $115, pointing to the company's aggressive spending and the AI division's sustained losses. The S&P 500, which has roughly $13 trillion in passive investment linked to it, confirmed it will not waive its eligibility rules: companies must have four consecutive profitable quarters and at least 12 months of trading history before the index considers them. SpaceX meets neither. The Nasdaq-100 did relax its rules, however, meaning SpaceX can join that index within 15 trading days of its debut. If you hold a Nasdaq-tracking fund, you may own SPCX already without having done anything.
Anthropicfiled its confidential IPO paperwork on June 1, ahead of SpaceX's debut. It is targeting an October listing at a valuation near $965 billion. The revenue numbers it has shared with investors are striking: $4.8 billion in the first quarter of 2026, projected to reach $10.9 billion in the current quarter, a doubling in a single quarter. That would make it the first AI lab to post a quarterly operating profit, at $559 million. Anthropic itself has been clear that this profit is temporary. Heavier spending on computing infrastructure in the second half of the year will push it back into losses. Its longer-term internal targets, reportedly shared with investors, are $70 billion in annual revenue and $17 billion in cash flow by 2028.
Anthropichas also been in an open dispute with the US Defense Department since January. The Pentagon demanded Anthropic remove guardrails from Claude that prevent its use for autonomous weapons and domestic mass surveillance. Anthropic refused. The Pentagon labeled it a "supply chain risk," a designation typically used for companies from hostile foreign countries, and ordered military contractors to stop using Claude. A federal judge blocked most of that designation, calling it punitive. Despite the official ban, the NSA is reportedly using Mythos, Anthropic's most capable model, for cyber defense work. Mythos is so powerful that Anthropic has declined to release it to the public at all: it is accessible only to around 50 security organizations through a controlled program called Project Glasswing.
OpenAI filed on June 8 and announced it directly: "We recently submitted a confidential S-1. We expect it to leak so we're just announcing it." The company targets a September listing and was last valued at $852 billion privately. It generates roughly $2 billion in monthly revenue. It also spends $1.22 for every dollar it earns and does not expect to reach profitability until 2029 or 2030.
Three things are worth holding on to as a business operator. First, all three of these companies will face quarterly public scrutiny for the first time once they list. The pricing, reliability, and feature decisions of the AI tools you use will increasingly be shaped by what shareholders want to see every 90 days. Second, Anthropic's current revenue trajectory confirms that demand for enterprise AI tools is real and accelerating, not just projected. Third, the Anthropic-Pentagon dispute shows that the companies building these tools have their own limits on what they will allow. For any organization using AI in sensitive or regulated work, understanding those limits is not optional.