Archive monthly-report-06-07-2026
All ReportsDaily ReportOptions
IPO ANALYSIS

SpaceX IPO • The Revised Read (June 7th 2026)

Generated Sunday, June 7, 2026 · Revised since the May 30 edition

Author Claude
Published 7 June 2026
Read ~14 min
Raw File monthly-report.md

The crowd that cannot buy is pricing $2.4 trillion. The company is selling at $135 a share. The gap between them is the entire trade.

Bottom Line Up Front

A week ago we framed SpaceX as three companies in one wrapper, priced for a future not yet built. Nothing in that thesis is wrong, but four things changed the picture between May 30 and today. The IPO went to a fixed price of $135 per share instead of a range, locking the deal at roughly $1.75 trillion with a $75 billion raise, trading expected June 12 on Nasdaq under SPCX (Reuters/Capital.com, June 3; TechCrunch, June 5). Google signed a $920 million per month compute deal, validating the AI revenue story (TechCrunch, June 5). S&P Global refused to fast-track SpaceX into the S&P 500, while Nasdaq and FTSE Russell did bend their rules. And Anthropic filed to go public, confirming the deal pipeline is real. The net effect: the bull case got a little more concrete, the bear case did not go away, and the single most tradeable fact is now an index mechanic, not a story about Mars.

Fact-Check — The Money Gang Group Chat

You asked me to check every claim. Here is the scorecard. Most of the chat was directionally smart with a few hard errors.

Claim in the chat Verdict Reality
IPOs let insiders dump pre-IPO shares on retail, "liquidity trap" Mostly true in spirit Lockups, underwriter incentives, and the historical underperformance of mega-IPOs all support this. "Guaranteed Great Depression" is hyperbole.
Index rules were fast-tracked to add SpaceX even though it is cash-flow negative Partly true Nasdaq-100 (15 trading days) and FTSE Russell (5 days) changed rules. S&P 500 did not.
"S&P 500 said nah you ain't getting in" True On June 4 S&P Dow Jones kept its profitability and float rules. SpaceX is not eligible until ~mid-2027.
Trading at ~$195 on Hyperliquid Roughly true The SPCX-USDC synthetic perp has traded $200 to $203 ($2.4T implied). It is a leveraged derivative, not the stock, and the IPO prices at $135.
$225B compute deal with Google False The Google deal is ~$30B total ($920M/month, Oct 2026–June 2029).
Signed a compute deal with Anthropic True $1.25B/month through 2029, roughly $15B/year.
Anthropic paying SpaceX $15B/year (the Reddit post) True $1.25B/month x 12 = $15B/year.
Colossus 1 is the largest supercomputer cluster in the world Marketing, contestable Colossus (now under SpaceX via the xAI merger) is among the largest single AI training clusters. Google is the largest single owner of AI compute overall.
Built the largest cluster in under 2 years / 122 days Roughly true xAI's initial Colossus build was famously ~122 days.
Tesla trades at ~200x forward earnings True ~213x forward P/E as of June 5 (GuruFocus).
$7T in AI data center spending by 2030 True McKinsey estimates ~$6.7T cumulative data-center capex by 2030, ~$5.2T AI-specific.
50% of retirement accounts are a bet on AI Exaggeration of a real risk The Magnificent 7 are ~35% of the S&P 500. Concentration is real, the 50% figure is loose.
50% of spending is driven by the top 10% of earners Roughly true Moody's Analytics has pegged the top 10% at roughly half of US consumer spending.
Anthropic reached recursive self-improvement and issued a warning Unverified Anthropic did file to IPO on June 1. I could not substantiate any official "recursive self-improvement" warning. Treat as rumor.
Data centers in space skip liquid cooling because space is cold Misleading physics Space is a near-vacuum. With no air there is no convection, so heat only sheds by radiation, which is slow. Orbital data centers have a harder cooling problem and need large radiators. Google and SpaceX are exploring orbital compute, but cooling is a challenge, not a freebie.

The one structural insight the chat got right and most retail misses: xAI is now part of SpaceX (merger effective February 2, 2026). That is why "SpaceX" owns Colossus and why Google and Anthropic pay SpaceX, not xAI, for GPUs. You are no longer buying a clean rocket company.

The Revised Theory

Last week's report leaned cautious on the grounds that the valuation prices execution not yet delivered. That still holds. But the past week added evidence that the AI-infrastructure leg is becoming a real cash business, not just a slide.

The Google deal is the tell. Google is, by most estimates, the largest single owner of AI compute on earth, and it still agreed to pay SpaceX $920M/month for roughly 110,000 NVIDIA GPUs from October 2026 through June 2029 (TechCrunch, June 5). When the most compute-rich company in the world rents capacity from you, your data-center segment is not vaporware. Stack that on the Anthropic deal ($1.25B/month, all of Colossus 1's available compute) and SpaceX has roughly $75B of contracted compute revenue across the two agreements over their terms, with cancellation clauses after December 31, 2026.

So the revised theory is this. SpaceX is converting from "rocket company with an expensive AI hobby" into a vertically integrated compute-and-connectivity utility: Starlink throws off cash, launch lowers the cost of everything downstream, and the AI segment is now signing nine and ten-figure contracts. The orbital-data-center pitch remains the long-pole bet, and the cooling physics are genuinely hard, but the near-term AI revenue is no longer hypothetical. The price still asks a lot. The story is just less of a leap of faith than it was seven days ago.

What You Are Actually Buying

Three engines, recast into one set of consolidated financials after the xAI merger.

Starlink, the cash machine. 10.3 million subscribers as of March 31, 2026, doubled in a year, roughly $11.4B revenue and $4.4B operating income in 2025 (prospectus via Yahoo/Motley Fool). This is the anchor and the reason the rest is plausible.

Launch, the moat. Reusable Falcon 9 is the workhorse. The forward story rests on Starship, which had completed ~12 flight tests but not delivered a commercial payload as of the filing. Payload delivery is targeted for H2 2026 (SpaceXChart). The valuation leans on a vehicle that has not yet done its commercial job.

AI, the cash furnace. xAI/Grok/X burns near $1B/month against modest revenue (~$500M ARR), and the consolidated entity reported a $4.28B GAAP loss in Q1 2026 (SpotGamma). The Google and Anthropic deals are the offset that could turn this leg cash-positive over the next two years if delivery hits.

At $1.75T against roughly $15-16B combined revenue, you are paying about 100x trailing sales. For context, even the richest mega-cap software names trade at a fraction of that. The bull rebuttal is that trailing multiples are the wrong lens for three fast-compounding businesses with an enormous terminal market. That may be true, but it means the valuation is a bet on execution, not a multiple of what exists.

The Bull Case

The Bear Case

Where Wall Street Is Pricing It

The sell-side read is more measured than the tape. The deal priced at a fixed $135 per share rather than letting a bookbuild range float up, which is itself a signal of disciplined demand management by the underwriters (Goldman Sachs, Morgan Stanley, BofA, Citigroup, J.P. Morgan). At $135 and ~13 billion fully diluted shares, that is the ~$1.75T mark.

The most important institutional signal of the week was a negative one: S&P Dow Jones refused to bend. A strategist quoted by Bloomberg put it plainly, that making exceptions for very large, long-private, still-unprofitable companies did not make sense. That is Wall Street's gatekeeper telling you the fundamentals do not yet clear its bar. Institutions can buy the IPO, but the largest forced-demand pool in the world is not opening on schedule.

Where Retail Is Pricing It

Retail is euphoric, and it is pricing the deal above where it is actually selling. The Hyperliquid SPCX-USDC synthetic perp opened at $150 on May 18, spiked to $216, and has traded around $200 to $203, implying roughly $2.4 trillion (BeInCrypto; Yahoo). That is a ~45% premium to the $135 IPO price. Binance, OKX, Bitget, and BingX run similar contracts. Prediction markets show high confidence the deal happens (Kalshi ~76% before September 1) but only coin-flip odds on the $2.4T valuation holding (Polymarket ~50/50).

The tell from last week still stands and is now sharper: enthusiasm is loudest exactly where people cannot buy real shares (synthetic crypto, prediction markets) and most measured where allocation is real (institutional desks). A synthetic long at $200 is underwater on day one if SPCX opens anywhere near its $135 deal price. The contract already had a 45% flash crash on May 28 that liquidated retail leverage. This is a price signal with no order book behind it.

The Index Mechanic — The Most Tradeable Fact

Forget Mars for a moment. The cleanest, most identifiable catalyst is mechanical.

This split is the whole near-term game. If SpaceX enters the Nasdaq-100 around late June or early July, passive QQQ flows become forced buyers into a tiny float. That is the strongest argument for a short-term bullish or day-trade lean. The catch: it is well telegraphed, so a good chunk is already in the synthetic price.

How Mega-IPOs Historically Go

The historical base rate is unkind, and the pattern is consistent (Motley Fool; Reuters/Gulf News; DealRoom).

The mechanism is structural: underwriters underprice to reward big clients, who flip for a quick profit, while retail buys the headline pop and holds the bag through the drawdown. Your friends' "liquidity trap" instinct is, in aggregate, supported by the data.

The Musk Exception

Now the other side, because Musk names break base rates. Tesla IPO'd in 2010 at a split-adjusted ~$1.30 and has been called "overvalued" continuously for fifteen years while compounding to a ~$213x forward P/E and a multi-hundred-billion-dollar company. The market has paid a structural premium for Musk execution that fundamentals never justified on paper, and shorting that premium has been a graveyard.

There is also a real economic linkage now. Tesla disclosed $143M in vehicle sales to SpaceX, and the market increasingly treats the two as one organism (BitMEX). The reported, unconfirmed SpaceX-Tesla merger chatter sits on top of that. The lesson is not "buy at any price." It is that a permanent-bear stance on a Musk vehicle has been wrong often enough that you should size a short thesis humbly.

Macro Backdrop (from the June 6 report)

The IPO lands into a jumpy tape. SPY closed June 5 at 737.55, sitting exactly on its gamma flip at 737, with dealers positioned to amplify moves in either direction (net gamma ~ -$579M, regime unstable). The VIX jumped about a third to just above 20 on a hot May jobs report stacked on a Broadcom miss. Crucially, credit never flinched (high-yield spreads ~272 bps, historically tight), which says the selloff was positioning and rates, not a solvency scare. Our composite read was 3.67/7, Neutral-Fear.

Two binary events bracket the listing: May CPI on June 10 and the FOMC plus dot plot on June 17, with quad-witching June 19. SPCX debuts June 12, between them. That means the IPO prints into a window engineered for outsized one-day moves. A cool CPI is the most bullish thing available; a hot one revives the stagflation trade and the rate-driven tech multiple reset. Plan for volatility that has nothing to do with SpaceX itself.

My Honest Take — Sit Out, Invest, or Day-Trade

You said this report is to decide between sitting out, investing, or day-trading. Here is my read, framed as analysis, not advice.

The asset quality is genuinely high and the entry price is genuinely rich. Both are true at once. Starlink alone is a wonderful business. You are being asked to pay ~100x sales to also own a $1B/month cash furnace and a roster of milestones not yet hit, into a fragile macro tape, with a tiny float and a synthetic crowd already pricing 45% above the deal.

If you are tempted to invest for real: the discipline is to separate the company (high quality) from the entry (unknown until it trades, rich at target). Mega-IPOs are usually cheaper six to twelve months later. A patient entry after the first SEC-filed quarter (early November 2026) gives you real numbers instead of a prospectus. If you want exposure now without the day-one casino, the XOVR ETF holds SpaceX via an SPV (reportedly 40%+ as of April) as a diluted proxy. Dollar-cost in. Do not chase the open.

If you want to day-trade it (high risk): the Nasdaq-100 forced-buying window (~15 trading days after listing) is the cleanest catalyst on the board, and the 3-5% float can squeeze. But expect violent two-way moves, use limit orders never market orders at the open, define your risk, and respect that synthetic longs at ~$200 are underwater versus a $135 deal, so a first-day fade is a live scenario. This is a trade, not a marriage.

If you sit it out: that is the most defensible default, not a cop-out. You avoid the low-float lottery, you keep powder for a likely better entry, and history says the patient buyer of mega-IPOs usually wins. The one risk you accept is that this is a Musk name, and Musk names have punished patience before.

My personal lean if I had to commit to one sentence: I would treat day one as a trade, not an investment, and if I wanted to own it for real I would wait for the float to grow and the first public quarter to print. The quality is not the question. The price and the timing are.

Other Threads Worth Pulling

This is analysis, not investment advice. Figures are drawn from the May 20 S-1, the June 5 SEC free-writing prospectus, and contemporaneous reporting through June 7, 2026. Several financials are estimates and may be revised. Verify against the final prospectus before acting.

Sources