OpenAI Starts Trillion-Dollar IPO Collision.

Mazen Al Sabbagh

— 6 min read
Share

TL;DR

  • OpenAI's IPO Filing: Expectation of a confidential SEC prospectus filing today, targeting a Q4 2026 listing at up to $1 trillion, positioning it for a significant market event alongside SpaceX and Anthropic.
  • AI Mathematical Breakthrough: An unreleased OpenAI reasoning model autonomously resolved an 80-year-old Erdos conjecture in discrete geometry, marking a precedent for AI-driven mathematical discovery.
  • Grok's Market Underperformance: A Reuters investigation highlights Grok's minimal traction within the US government, undermining a core AI growth narrative for SpaceX's projected $1.75 trillion IPO.
  • Samsung Labor Resolution: Samsung's 89,000 union members commenced voting on a provisional wage agreement, with ratification anticipated by May 27, averting an extended strike.
  • Nvidia's Valuation Drift: Nvidia shares concluded Friday down 1.8% at $219.51, extending their post-earnings decline amidst sustained "priced for perfection" market dynamics, despite strong financials.

Lead Story: OpenAI Starts Trillion-Dollar IPO Collision

The AI sector is poised for its most consequential price-discovery event. OpenAI is expected to confidentially file its IPO prospectus with the SEC as early as today, according to multiple outlets. Goldman Sachs and Morgan Stanley lead the deal, with JPMorgan Chase also involved, aiming for a public listing between Labor Day and Thanksgiving 2026 at an $850 billion to $1 trillion valuation.

This timing is strategically calculated. SpaceX publicly filed its S-1 on Wednesday, targeting a June 12 Nasdaq debut at $1.75 trillion — projected as the largest IPO in history. Anthropic has communicated a willingness for a public listing by October. Both OpenAI and SpaceX share lead banks, and Axios reported the filing sequence was intended to diminish SpaceX's market entrance. The Musk-Altman rivalry, recently concluded in court, now manifests on Wall Street.

However, their financial narratives diverge. Anthropic forecasts $10.9 billion in Q2 revenue and its inaugural operating profit. SpaceX disclosed $4.3 billion in Q1 net losses, with xAI consuming 76% of its capital expenditure. OpenAI's Q1 operating margin registered at a negative 122%, and CFO Sarah Friar has internally cautioned about potential inability to cover compute contracts if revenue growth falters. These concurrent filings will compel public markets to value fundamentally distinct AI business models, establishing benchmarks that will reshape sector architecture.

In Other News

An OpenAI model autonomously disproves an 80-year-old math conjecture. OpenAI announced Tuesday that an unreleased general-purpose reasoning model independently disproved a central conjecture in discrete geometry, originally posed by Paul Erdos in 1946. For nearly eight decades, the problem — maximizing unit-distance pairs among points in a plane — was assumed largely resolved by square grid constructions. The model discovered a new family of constructions utilizing algebraic number theory, yielding a polynomial improvement. Fields Medalist Tim Gowers applauded it as "a milestone in AI mathematics", with Princeton's Noga Alon, Melanie Wood, and Thomas Bloom independently verifying the result. Princeton mathematician Will Sawin further refined the construction. The unexpected bridge between algebraic number theory and discrete geometry transcends mere AI benchmarking; it opens new avenues for human mathematical inquiry. While full journal peer review is pending, verification by leading researchers lends substantial credibility.

Reuters investigation finds Grok is a flop with the US government. A Reuters exclusive published Thursday revealed xAI's chatbot has achieved negligible penetration in federal agencies, undermining the AI growth narrative central to SpaceX's IPO. Documented federal AI use cases show only 3 references to Grok, against 234 for OpenAI products and 33 for Google. Enterprise usage data from Netskope positions Grok at just 2 per 1,000 enterprise users — a decline from its peak of 5 — with those users spending less than half the time on Grok compared to ChatGPT. Last month, xAI failed in its bid to develop a Grok-powered product for the Department of Veterans Affairs due to unmet requirements. This report arrives at a critical juncture for SpaceX, whose S-1 projects a $28.5 trillion total addressable market for AI services while xAI sustains $6.4 billion in quarterly burn.

Samsung's 89,000 union members begin voting on the tentative wage deal. Electronic voting commenced Friday at 2:12 p.m. local time following a brief server-related delay, continuing until May 27. The agreement includes a 6.2% average salary increase and a 10.5% profit-linked stock bonus for the chip division, with certain memory workers eligible for up to $416,000 — although bonuses are vested in company stock over a minimum of 10 years and contingent on the division achieving 200 trillion won in annual operating profit. Reuters characterized the deal as favorable to Samsung, noting the less generous bonus structure compared to SK Hynix's uncapped 10% profit-sharing. Union leader Choi Seung-ho anticipates ratification but acknowledged discontent among foundry and LSI workers, whose compensation will be substantially less than memory division colleagues. Samsung shares surged 8.5% to a record high on Thursday.

X / Social Pulse

The mathematical breakthrough dominated researcher discourse, with OpenAI's assertion that its model "discovered an entirely new family of constructions" fueling sharp debate regarding genuine scientific discovery versus sophisticated pattern-matching at scale. Musk responded to the trial verdict, posting on X that "the judge & jury never actually ruled on the merits of the case, just on a calendar technicality," accusing Altman and Brockman of "stealing a charity." Altman, conversely, made headlines by offering YC startups $2 million in free AI tokens for using OpenAI products—a move critics labeled desperate and supporters called astute distribution warfare. The Reuters Grok investigation provided ample ammunition for SpaceX skeptics; the "3 versus 234" federal use case ratio circulated across financial Twitter as shorthand for the chasm between xAI's ambitions and its adoption. Politico's revelation that David Sacks intervened to nullify the AI executive order prompted pointed commentary on the Silicon Valley-White House revolving door.

One to Watch

The Sacks intervention that killed Trump's AI order. Politico reported Thursday night that former White House AI czar David Sacks raised last-minute industry concerns, leading Trump to withdraw his AI executive order hours before its signing. Sacks warned that the voluntary 90-day pre-release review framework could eventually become mandatory—a apprehension shared by several tech companies who dispatched lower-level executives rather than CEOs to what was intended as a marquee event. This collapse creates a federal AI regulation vacuum precisely as California intensifies its legislative efforts: Newsom's workforce protection order signed the same day explores universal basic capital and AI layoff tracking, the state Senate recently passed the No Robo Bosses Act, and the DOJ has joined xAI in suing Colorado over its AI anti-discrimination law. The question is no longer if AI will be regulated, but whether states will establish the regulatory framework while Washington debates the necessity of any rules at all.

Quick Hits

  • Nvidia shares fell 1.8% Friday to $219.51 — The "priced for perfection" dynamic persists as Nvidia's stock continues its post-earnings slide, now 7.2% below its May 14 peak, despite strong results, marking its sixth consecutive post-earnings decline.
  • Standard Chartered is cutting 7,000+ jobs, explicitly citing AI to replace "lower-value human capital" — A candid acknowledgment of AI-driven workforce displacement, providing a stark framing for its impact on human capital deployment.
  • Zoom reported Q1 revenue of $1.24B (up 5.5% YoY) with AI Companion paid users surging 184%, sending shares up 8% after hours — Zoom's AI integration demonstrates a viable path to revenue growth, defying prior market skepticism regarding its long-term relevance.
  • Smart ring maker Oura filed confidentially for an IPO at an $11B+ valuation, joining the 2026 filing wave alongside SpaceX, OpenAI, Discord, Kraken, and Strava — Oura's confidential IPO filing at an $11B+ valuation signals the expansion of the 2026 public market surge beyond core AI, into adjacent tech sectors.
  • Chinese AI models now account for 60% of all usage on OpenRouter, according to BuildFastWithAI — This market share shift on OpenRouter challenges the entrenched narrative of singular US dominance in frontier AI development, underscoring evolving global market dynamics.

This week marked the definitive transition of AI from a technology narrative to a financial markets story. Three entities, commanding an aggregate valuation of $3.5 trillion, are racing toward public listings within months, forcing an unprecedented level of transparency via S-1 disclosures. OpenAI's negative operating margins. SpaceX's xAI losses. Anthropic's $45 billion compute bill. These figures are now public, and the chasm between ambition and economic reality is starkly visible. Concurrently, an AI model quietly resolved an 80-year-old mathematical problem, a potent reminder that beneath the market theater, the underlying technology continues to advance in ways that dwarf any stock ticker.

Sources

Lock in. M. mazen@thorterminal.com

Read more