On the morning of June 12, 2026, Space Exploration Technologies Corp. began trading on the Nasdaq under the ticker SPCX. The night before, underwriters had priced 555.6 million shares at $135 apiece, raising roughly $75 billion. When the overallotment option was exercised in full a few days later, the final tally came to $85.7 billion. That is close to triple what Saudi Aramco raised in 2019, and it made SpaceX the largest initial public offering in the history of capital markets by a margin nobody is likely to challenge this decade.

The first trade came in near $150. By the close, the stock was at $161, up 19 percent, and SpaceX was worth about $2.1 trillion. It kept going. The following Monday added another 20 percent, and on Tuesday, June 16, SPCX touched $225.64 intraday. Four trading days after the bell, the company had gained roughly two-thirds on its offer price and briefly carried a market capitalization north of $2.5 trillion.

Then it went the other way, and it did not stop. On July 15 the stock closed below its $135 offer price for the first time. On July 23 it printed an all-time low of $110.85. It closed the week at $115.07.

None of this is especially mysterious, and it is not a referendum on whether Falcon 9 works. Three things happened at once. The prospectus told investors something they had spent five years assuming was untrue, only about four percent of the company was actually available to trade, and the calendar for fixing that second problem begins in the first week of August.

The Prospectus Broke the Story Everyone Had Been Telling

For most of the last decade, the received wisdom about SpaceX’s finances came from leaks, secondary-market chatter, and one genuinely useful document, the annual filing from Starlink’s Netherlands subsidiary. In January 2026, Reuters reported that SpaceX had generated roughly $8 billion in EBITDA on $15 to $16 billion of revenue in 2025. That figure circulated everywhere, including in our own coverage of SpaceX’s ownership structure. It described a company printing money.

The S-1 that SpaceX filed on May 20, 2026, described something else. Consolidated 2025 revenue was $18.7 billion, higher than the estimates, because the February 2026 acquisition of xAI pulled a new business line into the accounts. Consolidated operating loss was $2.6 billion. Net loss was $4.937 billion.

Starlink is an outstanding business. It cleared $11.4 billion of revenue in 2025, 61 percent of the company total, and threw off $4.42 billion in segment operating profit. The launch business is smaller and also profitable. The thing dragging the consolidated line into the red is the AI segment, which lost $6.36 billion on $3.2 billion of revenue in its first year inside the house.

The capital intensity is the part that alarmed people. Of SpaceX’s $20.7 billion in 2025 capital expenditure, $12.7 billion went to xAI. In the first quarter of 2026 alone, the AI segment consumed $7.72 billion in capex, more than Space and Connectivity combined, on the way to a $4.276 billion quarterly net loss. Long-term debt stood at $29.1 billion as of March 2026. Investors who had spent years waiting to buy a rocket company found themselves reading a prospectus for a data-center buildout with a rocket company attached to fund it.

Four Percent of a Company Is Not a Market

The second problem is mechanical, and it made the first two weeks of trading almost meaningless as a price signal.

SpaceX floated a very large dollar amount and a very small fraction of itself. The company carries roughly 13 billion shares outstanding after a 5-for-1 pre-IPO split, so the 555.6 million sold in the offering worked out to about 4 percent of the company. The other 95-odd percent sat under lockup. At the same time, the underwriting syndicate did something unusual with the allocation, steering an initial 20 percent of the deal to retail investors, worth about $15 billion on its own and three to six times a typical individual allocation. Order books reportedly reached about $150 billion against a $75 billion raise, and the retail tranche was exhausted before the stock ever opened.

That combination produces exactly the price action SpaceX got. Enormous demand meeting a deliberately starved supply drove the stock to $225.64, then the demand thinned out and there was nothing underneath it.

The unwind started in the third week of June. After a 5 percent drop on Wednesday and a 3.6 percent drop on Thursday, SPCX fell 16.4 percent in a single session on Monday, June 22, the same day the company launched its first bond offering, a $20 billion investment-grade issue meant to refinance the bridge loan that had funded the xAI acquisition. All three major agencies had rated the notes investment grade, at Baa1, BBB+, and BBB. The market read the deal less as a vote of confidence than as confirmation that the AI segment’s cash requirements were going to be met by the balance sheet.

August 6 Is the Date That Matters

SpaceX did not write a single 180-day lockup. It wrote a staircase, and the first step is triggered by the company’s first earnings report.

SpaceX will post second-quarter results after the close on August 4, with Musk on the 4:30 p.m. ET call. Two full trading days later, on August 6, insiders can begin selling as much as 20 percent of their restricted holdings, a block that runs to roughly 911.5 million shares. Tranches of about 7 percent follow on August 21 and September 10. A larger release of around 28 percent is tied to the third-quarter report, and everything still locked comes free on December 8, the 180-day mark, at which point something like 40 percent of the company is tradable.

Musk himself is on a 366-day lockup and cannot sell until roughly June 2027, which removes the single largest overhang from this year’s calendar but does nothing about the other 240-odd shareholders on the pre-IPO cap table.

Wall Street Cannot Agree Within a Factor of Four

Twenty-five days after the debut, when the underwriters’ research quiet period expired, 18 banks published price targets. The spread is remarkable. Raymond James came in at $800, betting on weekly Starship flights by 2028. Stifel published $190. Morgan Stanley sat at $300, Goldman Sachs at $205, and four firms including JPMorgan and Deutsche Bank clustered on a $225 median. William Blair issued an outperform rating and declined to publish a target at all.

Nine of the 18 clustered between $200 and $225, which says less about independent modeling than about the difficulty of the exercise. The two most bearish notes came later and from firms without underwriting economics in the deal. MoffettNathanson’s Julie Zhu initiated at Neutral with a $131 target, writing that SpaceX lacks a credible financial model to support a valuation near $2 trillion. On July 23, HSBC’s Nicolas Cote-Colisson started coverage at Hold with a $115 target, applying what CNBC described as a two-times innovation premium for Musk’s track record and still concluding the shares were fully valued.

HSBC’s target turned out to be the accurate one. SPCX closed at $115.07 on July 24 for a market capitalization of $1.51 trillion, within a rounding error of the number Cote-Colisson had published the day before.

The bear case is a sum-of-the-parts exercise. Value Starlink and the launch business on comparable multiples for infrastructure and connectivity assets, assign the AI segment something defensible for a unit losing $6 billion a year, and the total comes out near $900 billion, or roughly $69 a share. Even after the drawdown, the market is paying a substantial premium over that for xAI. At $1.51 trillion against $18.7 billion of 2025 revenue, SPCX trades around 80 times trailing sales. The price-to-earnings field on its quote page is blank, because there are no earnings.

The Rockets Are Doing Fine

The operating business had a good month, which is easy to lose track of while watching the ticker.

Starship Flight 13 aborted at T-0 on July 16 when four Raptors on the Super Heavy booster failed to ignite. The automated system did what it is supposed to do. Two engines were swapped, and the vehicle flew on July 24. All 33 booster engines lit, the ship deployed 20 operational Starlink V3 satellites, relit a Raptor in space, and made what observers called the softest splashdown the program has recorded. The booster missed its own soft water landing when engines failed to relight on descent, which is the one real blemish. Each V3 satellite carries roughly 1 Tbps of downlink capacity, about ten times a V2, and 160 Gbps up, a 22-fold improvement. That deployment is the first time Starship has done real work rather than carry mass simulators.

Starlink itself passed 10.3 million subscribers across 164 countries by March 2026, up from 8.9 million at the end of 2025 and 2.3 million at the end of 2023. The subscriber curve is not the concern. Average revenue per user has fallen from about $99 a month in 2023 to $66 in the first quarter of 2026 as growth shifted to lower-income markets, which is a deliberate choice and a defensible one, but it means the revenue line grows slower than the user line.

SpaceX has also quietly stopped accepting new commercial Falcon 9 and rideshare bookings beyond 2028 and begun winding down production of some expendable Falcon components, while continuing to fly NASA and Pentagon manifests. That is a company betting its own manifest on Starship reaching operational cadence. Our daily SpaceX X Report tracks each of those flights as they happen.

What August 4 Actually Decides

The first earnings call will not resolve the valuation argument, but it will settle three narrower questions that the stock is currently guessing at.

The first is Starlink subscriber momentum. Analysts are looking for 11.5 million or better, and a miss there attacks the only segment nobody disputes. The second is ARPU. If the decline from $99 to $66 continues at that slope, connectivity revenue growth decelerates regardless of how many dishes ship. The third is AI capex guidance. A $7.7 billion quarter annualizes to more than $30 billion, against $12.7 billion for all of 2025, and management has not told public investors where that line stops.

Then, two days later, the selling window opens.

My read is that the drawdown is a repricing rather than a verdict. A company with 4 percent of its shares outstanding does not have a real price, it has an auction result, and the auction result was $225.64. The S-1 numbers were public three weeks before the IPO priced, so nothing about the losses is new information. What changed is that enough shares are about to exist for the market to actually express an opinion, and the market is front-running its own supply. Expect the volatility to persist through the December 8 expiry rather than resolve in August.

The harder question is the one HSBC and MoffettNathanson are both circling. SpaceX is now two companies with opposite financial profiles stapled together by a merger that Musk approved with 93.6 percent of the Class B super-voting shares in his pocket. Starlink and launch generate cash. xAI consumes it faster than they generate it. Public shareholders bought Class A stock with no practical ability to change that arrangement, in a Texas-incorporated controlled company that requires a 3 percent stake to file a derivative suit and routes most disputes to arbitration. If the AI bet works, none of the governance matters. If it does not, the people who own the profitable half of this company have no mechanism to protect it. That is the actual risk in SPCX, and it will not be settled on August 4.

Frequently Asked Questions