SpaceX Files for SPCX, Cerebras Sets the Convergence Trade, and the Pre-IPO Market Stops Being a Forecast

This week SpaceX filed for what would be the largest stock-market debut in history. Reuters first reported the filing, and follow-up coverage from Barron's and CNBC put the contours at a roughly $1.75 trillion valuation with a potential raise near $75 billion, targeting a Nasdaq listing as early as June 12 under the ticker SPCX. Barron's framed the obvious risk for retail allocators — an opening-day cross at a trillion-dollar-plus float will be the most violently price-discovered event in modern IPO history. But the more interesting story for educated investors is what was happening in the pre-IPO window before the S-1 ever hit EDGAR.

For the first time, the synthetic pre-IPO market is producing a continuous, dollar-denominated, openly observable price on a company that has not yet listed. On May 18, the same day Reuters reported the SpaceX timetable, TradeXYZ opened a SpaceX market that has since traded between roughly $180 and $230, implying a fully-diluted valuation in the neighborhood of $2.5 trillion against the reported ~11.87 billion shares outstanding. Ventuals, the rival pre-IPO perpetual on Hyperliquid, prices it as a valuation unit rather than a share price. Both numbers sit materially above the formal IPO target. That gap is the trade. And four days earlier, with Cerebras, it worked.

The Cerebras Convergence Trade Just Validated the Mechanism

Cerebras Systems (CBRS) opened on the Nasdaq IPO Cross on May 14 at $350, after the company filed its S-1 on April 17 and revised its bookbuild range to $150-$160 from an initial $115-$125 on demand reported at more than 20x cover. Synthetic pre-IPO perps tracking CBRS had already been trading at a 77 percent premium to the high end of the formal bookbuild for days. In the final hour before the open, the VWAP of the CBRS pre-IPO perp sat just 1.2 percent above Nasdaq's eventual print. In the final minute, it was 2.8 percent above. The market processed over $200 million in pre-open notional volume at a 61.4 basis-point median spread — numbers that would have been impossible a year ago for a contract with no public reference.

This is the part that matters for anyone studying private-market price discovery. A cash-settled synthetic perpetual, with no claim on the underlying shares and no external oracle feed, anticipated the public open with what one observer called "institutional-grade precision." Five years ago, the only way to express a directional view on a pre-IPO name was through a secondary platform like Forge Global, EquityZen, or Hiive, paying a 5-15 percent bid-ask spread on lots that typically required $25,000 minimums and 30-90 days to close. The synthetic perp model didn't replace those venues; it added a second, faster, continuously-priced layer on top of them.

Worth noting: the Cerebras post-IPO trade was not as kind. Open interest in the CBRS perp collapsed from $57.77 million at peak demand to about $48 million, and 24-hour volume fell from $280 million to roughly $53 million. Anyone who held the perp through the open absorbed a roughly 29 percent peak-to-trough drawdown on what was technically a successful convergence. The pre-IPO market got the open right and was still a punishing place to be long after the bell.

Anthropic's Implied Trillion-Dollar Valuation Is the Other Half of the Story

While SpaceX and Cerebras dominate the IPO-imminent end of the spectrum, Anthropic represents the other extreme: a company that may stay private for years and is being priced in real time across at least four different venues that disagree by hundreds of billions of dollars. The last formal funding round, a Series G closed in February 2026, put Anthropic at $380 billion. By mid-April, Forge Global secondary trades implied a valuation near $1 trillion. Tokenized PreStocks priced it as high as $1.6 trillion. Ventuals briefly marked Anthropic above OpenAI at an implied $846 billion on April 11. Hendrik Jordaan's mid-May secondary snapshot put it at $1,449.32 per share, the highest per-share mark of any major private name he tracks.

And on May 21, news broke that Anthropic had signed a $45 billion, three-year compute contract with SpaceX — roughly $1.25 billion per month through May 2029. That single contract reframes both companies' pre-IPO marks simultaneously. SpaceX gets a contracted revenue line the IPO bookbuild can underwrite. Anthropic prints a new floor on its capacity requirements. The Forge-implied trillion-dollar mark stops looking aspirational.

Polymarket Plus Nasdaq Private Market Adds a Prediction Layer

The third structural development of the week is institutional. Polymarket and Nasdaq Private Market announced an exclusive partnership in which NPM provides institutional-grade transaction and pricing data to settle yes-or-no contracts on private company milestones. The initial market list reads like the AdValorem watchlist: OpenAI, Anthropic, Stripe, Databricks, Anduril, Neuralink, and SpaceX. Contracts will cover valuation thresholds, IPO timing windows, and secondary-market share prices.

What's interesting about the design is that it gives institutional allocators a way to express probability-weighted views on private companies without ever touching a share, a special-purpose vehicle, or a synthetic perp. A family office that already runs a Polymarket book on macro or election events can now run a book on whether SpaceX prices above $1.5 trillion on first-day close. The data flow runs in the opposite direction too: NPM's settlement data, previously available only to subscribers and counterparties on its ATS, is now anchoring a publicly-observable prediction market. That is a meaningful transparency step for a venue that has historically been one of the more opaque corners of late-stage venture.

What This Means for Educated Allocators

Three observations are emerging from the May 2026 data:

• The pre-IPO discount is compressing. Between Forge, EquityZen, Hiive, synthetic perps, and now Polymarket-NPM, the gap between a private mark and an implied public open is narrower and faster-moving than at any point since the 2014 secondary-market boom. The arbitrage window that used to last quarters now lasts weeks.
• Opening-day risk is the new pre-IPO risk. Cerebras showed that getting the open right is not the same as getting the post-IPO trade right. A 29 percent drawdown from a "successful" convergence is a teaching moment for anyone planning to hold SPCX through the bell.
• Price discovery is now multi-venue. A serious view on SpaceX in May 2026 requires reading Forge secondary prints, the SPCX synthetic perp on TradeXYZ, the implied valuation on Ventuals, the Polymarket-NPM milestone contracts, and the actual S-1 disclosures. No single venue is dispositive. The investors who do this well are the ones who can read all five.

This is why we publish daily research at gp.advalorem.io and frame our Pre-IPO Markets vertical as education. The mechanics of how private companies are now priced — through hybrid oracles, prediction markets, synthetic perps, and the slower secondary venues underneath them — have moved far beyond what most allocators were taught in business school. The companies we follow most closely as research topics (the AI Robotics cluster, the Quantum Pre-IPO names like Quantinuum and IonQ, our Newchip Warrant Portfolio coverage, and the Accelerator Warrant Enforcement framework) all sit at points along this same private-to-public continuum. Understanding the venues is the precondition for understanding the marks.

Whether SPCX opens at $1.5 trillion, $2 trillion, or somewhere outside both numbers in mid-June, the more durable lesson from this week is that the pre-IPO market is no longer a forecast. It is a continuous, multi-venue, openly-observable price — and the educated allocators are already reading it.