SpaceX has lost half its value since June. What actually happened

1. The shape of the fall

The sequence matters more than the size, because the dates rule out several of the explanations being offered.

SpaceX priced at $135 on 11 June and began trading on 12 June, opening near $150. It peaked four days later at $225.64. By 23 June it had already fallen to $147.11. It joined the Nasdaq-100 on 7 July and fell that day. It closed below its offer price for the first time in mid-July, and bottomed at $107.01 on 28 July.

Read that back and one thing stands out. The top was set four days after listing, three weeks before index inclusion. The decline was already well established before any of the mechanical index buying arrived. Whatever drove the peak, it was not the index.

2. Why the early price was never very informative

The free float at listing was 638,888,888 shares, about 4.9 % of roughly 13 billion shares outstanding. One share in twenty could actually change hands.

A price set by one twentieth of a company meeting the largest retail audience any listing has ever had is not a considered valuation. It is a scarcity reading. The same arithmetic that let the price reach $2.66 trillion on modest volume let it fall just as fast when the marginal buyer stopped showing up.

3. Three things that changed underneath

Three developments, none of them about rockets, account for most of the decline.

The AI capital-expenditure trade derated. Through July the market turned against companies spending heavily on AI infrastructure without near-term revenue to show for it. SpaceX, whose orbital data-centre ambitions sit squarely in that category, was repriced alongside the group rather than on its own news. Morgan Stanley's Adam Jonas wrote to clients about the gap that had opened between sentiment and fundamentals.

The supply calendar came into view. As the August lock-up date approached, the market began pricing the arrival of shares that do not exist in the float today. This is ordinary and well documented, and it does not require anyone to have a negative view of the company.

The broad market turned. The 29 July session was ugly for everything: the Fed held rates with three officials dissenting in favour of a hike, the S&P 500 fell 1.52 %, the Nasdaq fell 1.74 %, and yields rose. A high-multiple, long-duration listing was never going to be spared.

4. The two dates in early August

4 August. SpaceX reports quarterly results for the first time as a public company. There is no track record of guidance, no history of how management frames its numbers, and no consensus built on prior prints. First reports are volatile for that reason alone, independently of what the numbers say.

6 August. Restrictions lift on 911.5 million shares, valued near $116 billion when Bloomberg reported the schedule on 21 July, and closer to $103 billion at the 29 July close. Set against a float of 639 million, that tranche alone more than doubles the shares able to trade. Further tranches follow through December, and estimates of the cumulative effect vary, with some analysts describing the eventual float expansion as a quadrupling.

Two points of precision are worth holding onto here. Unlocked shares are permission to sell, not an obligation, and a released tranche is not a forecast of selling. And the schedule has been public since the offering documents, so it is not new information, only newly imminent.

5. What has not changed

It is worth separating the share price from the operating record, because they have moved in opposite directions.

Starship Flight 13 flew on 24 July, deployed twenty Starlink V3 satellites, relit an engine in space and splashed down on target. It was not flawless: the booster failed to relight enough engines for its landing burn and came down hard, and the satellite deployment was suborbital, a verification step rather than an orbital delivery. Observers called it mostly successful, which is the accurate description.

Starlink received a conditional exemption on 27 July, valid to 1 February 2028, from the ban on foreign-made consumer routers. Amazon Eero, NetGear, Askey and Sercomm received comparable treatment, so this removes a supply-chain risk rather than conferring an advantage.

The orbital data-centre filing of January 2026, covering up to one million compute satellites, has been placed on an FCC fast-track with the environmental impact assessment waived. Atmospheric scientists have objected on ozone-chemistry grounds and astronomers on observation grounds. For scale, about 26,890 satellites have been placed in orbit in total since 1957.

6. What the market is watching next

Not predictions, just the observable questions that the next two weeks will answer.

1. 01How much of the unlocked stock actually reaches the market. The gap between eligible and sold is where the whole overhang question is settled. It will be visible in volume and in subsequent filings, not in advance.
2. 02Whether the first earnings report reframes the AI spending. The derating has been applied to the category. A first print is the company’s first chance to describe its own capital plan in its own numbers.
3. 03Whether index mechanics matter on the way down. A larger float changes index weightings. That works in both directions and is a technical factor, not a verdict on the business.
4. 04The regulatory track on orbital compute. A waived environmental assessment shifts the burden to challengers. Whether it survives challenge is a live question with a long timeline.

7. What this episode illustrates

We are not offering a view on where the share price goes, and nothing here is a recommendation to buy, sell or hold anything.

What the episode does illustrate is narrow and checkable. A float of 4.9 % made the opening months of trading close to uninterpretable as a valuation signal. The supply schedule that has dominated July was published in the offering documents before the first trade. The index-inclusion narrative that grew up afterwards does not fit the dates. And the operating record moved independently of all of it.

Float, index rules and lock-up calendars are all disclosed in advance. In this case they explained considerably more of the price path than anything that happened in Boca Chica.