The Announcement Outlasts the Holder

3 September 2026. Figures are as of the dates stated in the text.

Interest disclosure: T-SpaceX is a tokenized loan participation right providing economic exposure linked to SpaceX, issued through a dedicated issuing subsidiary, as are T-OpenAI and T-Kalshi. Following the SpaceX listing, the investment underlying T-SpaceX is being divested. Redemption opens only once a Redemption Start Date is announced, and as of 3 September 2026 none has been. Tessera's founder, and entities he controls, hold related exposure. This piece is about that company. Read what follows with that in mind.

On 25 August, SpaceX announced a spaceport on a former Exxon property in Vermilion Parish, Louisiana, ending what the company called a seven-year search for a site. SpaceX says it is investing $100 billion, with construction starting in 2027, and claims a first Starship launch in 2029. Louisiana Economic Development, the state agency, gives 2030 for initial operations rather than 2029.

Who Counts the Capex, published on 17 August, examined why a number like that is never marked against anyone. Incurred capital expenditure enters the cash flow statement and is re-examined every ninety days, while announced, non-binding capital expenditure creates no obligation, commits no period, and leaves no line item to record a shortfall if it never happens. That asymmetry remains the important one in this sector.

But that piece asked what happens to the number. It did not ask what happens to the people. Three years is a long time on a share register, and this particular register is turning over on a published schedule.

The Interval Before Anything Is Auditable

Take the dates at face value: announced in 2026, ground broken in 2027, first operational use in 2029, or 2030 on the state's figure. Nothing about the intention is auditable before 2027 at the earliest, and nothing about whether it produced anything is auditable before 2029. In the meantime the figure is available to be reasoned about, modeled, and priced. It is denominated in the same units as the $18,369 million of second-quarter capital expenditure the company reported on 4 August ($15,828 million of it in the AI segment, against quarterly revenue of $7,814 million), it arrives through the same channels, and it carries none of the same consequences.

That was the 17 August argument. The part it did not address concerns the register itself.

The Register Turns Over on a Published Schedule

SpaceX listed this year, and its shareholder base is rotating rather than settling, on dates that were published in advance. The schedule requires no inference. It is printed in the prospectus, in a table headed "Earliest Date Available for Sale in the Public Market", with every tranche stated as "up to" and every tranche excluding affiliates unless noted otherwise:

That is the 180-day lock-up. A second, longer schedule runs beneath it, and it is easy to miss:

Eligible is not the same as sold; Eligible Was Not Sold, published on 10 August, argued exactly that. But eligibility is the precondition, and here it required no forecasting. The tables were published in June, anyone can read them, and they run for some fourteen months.

The second row of the first table deserves a pause. Those 455.8 million shares released only if the close held at least 30% above the $135.00 offer price, which is $175.50, on five of the ten trading days ending on the earnings date. Price-gated early release is standard underwriter practice rather than anything unusual. It matters here because it makes explicit what the rest of the table leaves implicit: how much stock becomes sellable is not independent of where the stock trades, and the issuer set that dependency down in writing, in advance, months before any of it happened.

There is a demand side to this with a calendar of its own, governed by index inclusion rules. We have not been able to confirm those rules against the index methodologies themselves, so they are left out of this piece rather than described second-hand. The argument does not need them; the selling schedule alone is published, dated, and sufficient.

Three features of those tables matter more than the individual numbers.

It finishes, and long before 2029. The last tranche clears on the second trading day after second-quarter 2027 results, some fourteen months after the schedule began. The program announced on 25 August is not testable in any respect until 2027, and not operationally until 2029 or 2030. This is a concentrated rotation at the front of the project, completed roughly two years before the first Starship is meant to leave the pad, followed by silence.

The largest thing released is the founder's entire position. On 12 June 2027, up to 6.4 billion Class A shares held by Elon Musk become eligible, an order of magnitude larger than any other tranche and the last dated release in the schedule, though not the final one: a further 351.9 million, the remainder of the extended lock-up other than his, follows after second-quarter 2027 results. The extended lock-up is drawn to release everyone else's stock in staged blocks around his, with the final block falling after it.

And what comes off lock-up is insider stock: founders, employees, early investors. The June buyers were never locked up and can hold to 2029 if they choose, and index vehicles hold indefinitely by construction. The claim here is not that the shareholder base empties out, and we have no turnover data that would support such a claim. The narrower, checkable claim is that the cohort released on this schedule is the cohort closest to the announcement, the people who were inside the company while the seven-year site search was running, released on a schedule that completes two years before anyone can check it.

Voting Control Comes First

One objection arrives before the claim, and it is correct. Per the prospectus, Elon Musk holds approximately 82.4% of the voting power. On a register like that, no constituency holds anyone accountable to anything, whatever the lock-up calendar does; the forum described below as disappearing was never open in the first place. Anyone reaching for a single explanation of why a $100 billion announcement will never be examined should reach for that one. It is simpler, it is in the filing, and it is sufficient.

The claim is therefore put narrowly, as the thing dual-class control does not already explain. Voting control determines who can compel an answer. It says nothing about who is in a position to ask the question at all: who was holding when the number was announced, priced it into what they paid, and would notice if it quietly changed shape. That is a memory function rather than a governance function, and it sits with whoever happens to own the stock at the time. Dual-class control removes the vote; the rotation removes the memory. The second is the smaller effect, and the less examined one.

The two are not as separable as that sentence implies, though, and the schedule says so. The votes are attached to shares, and those shares have a release date of their own: 12 June 2027, when up to 6.4 billion of them become eligible. Control of this company, too, carries a date, and it is the largest single line in a table published in June 2026.

The Constituency Cost

The 17 August piece said announced spending carries no reporting cost. The rotation adds a second absence: an announced number also carries no constituency cost. By the time a long-dated announcement becomes checkable, the holders best placed to interrogate it, meaning the ones who were inside the company when it was made and the ones who bought at prices reflecting it, have had some fourteen months of scheduled exits, ending two years before the thing is testable. Nobody has to forget anything deliberately. The register simply is not the same register, and the schedule that changed it was fixed at listing, before the announcement existed.

That reframes the disclosure asymmetry. The announced column escapes measurement, and it also answers to a different population of owners than the reported column, separated by years. The reported column answers to whoever holds the stock this quarter. The announced column, if it is ever answered for at all, answers to whoever holds the stock in 2029.

This is what "an IPO is an exit, not an entry" looks like when followed past the slogan. The exit is more than a liquidity event for early holders. It is the point at which accountability for long-dated statements passes from the people who made them to the people who bought in afterwards.

What This Argument Does Not Claim

It alleges no design. Nobody schedules a lock-up calendar around a spaceport announcement; the lock-up dates were set at listing, and the index rules were written by third parties with no interest in either. The mechanism requires no one's intent. It is calendar arithmetic, and it would operate identically if every statement made were entirely accurate.

It does not claim the announcement is inflated. A launch complex of the description given would plausibly cost something in that region, and the company has completed large physical programs on schedules outside observers doubted.

It attributes no price move to any of this. The 17 August piece declined to do so for a much cleaner event, and nothing has changed.

And we do not know the phasing. If the $100 billion is committed in dated tranches in a filing, most of this weakens: the number enters the measurement regime, and the interval closes.

Two Strong Objections

Two things cut against this argument, and both are serious.

The rotation may not be an exit into a vacuum. The standard answer is that as insider shares become eligible and the float rises, index-tracking demand rises with it, making this a rotation between cohorts rather than a withdrawal of capital, with the incoming cohort buying at prices that already reflect the announcement rather than in ignorance of it. That answer rests on precisely the index mechanics this piece declined to describe second-hand, and they are not reintroduced here because they happen to arrive as an objection. Take it as a real one we cannot presently price.

And a company with genuinely long-horizon assets ought to make long-horizon statements. Rockets are not a quarterly business, and requiring firms to say nothing about 2029 until 2029 would produce worse-informed markets, not better ones. The trouble lies elsewhere: the horizon of the statement and the horizon of the shareholder have come apart, and only one of them is disclosed.

What Follows

Nothing here is a prediction, and none of it requires the spaceport to fail. It is a structural observation with a dull conclusion: in a company whose register turns over faster than its capital program completes, the reporting regime marks the quarter, and nobody marks the decade.

The private mark is set by the marginal buyer at the last round. The public mark is set by the marginal seller after the lock-up. What sits between them, in a case like this one, is a set of statements about the 2030s made to an audience that will have changed by the time they can be tested. That is a strange thing for the most heavily scrutinized part of the capital structure to be carrying, and a stranger one for nobody to be counting.

Sources: the SpaceX Louisiana spaceport announcement of 25 August 2026, per Sean O'Kane, TechCrunch, published 11:26 PDT that day and read 30 August 2026; the $100 billion figure and the 2027 construction start are SpaceX's own statements as reported there, the 2029 first launch is reported as the company's claim, and the end-2027 construction start with initial operations in 2030 is attributed to Louisiana Economic Development. The lock-up release schedules, both the 180-day schedule and the extended schedule running to the second trading day after second-quarter 2027 results, including the release on 12 June 2027 of up to 6.4 billion shares held by Elon Musk, his entire position, together with the 30%-above-offer price condition on the Additional Release Shares, the $135.00 initial public offering price, and the approximately 82.4% of voting power held by Elon Musk, are from the SpaceX final prospectus (Form 424B4) filed 12 June 2026, read from EDGAR 30 August 2026. Second-quarter capital expenditure of $18,369 million, of which $15,828 million in the AI segment, and revenue of $7,814 million are from SpaceX Form 10-Q for the quarter ended 30 June 2026, filed 4 August 2026, segment tables, read from EDGAR 30 August 2026. Figures are as of the dates given. The SpaceX figures above are taken from the filings themselves; no index-methodology claim is made in this piece.

This is market commentary, not investment advice. It is not a recommendation to acquire, hold or redeem any Tessera product, or to take or avoid exposure to any company mentioned.

T-Tokens are tokenized loan participation rights, not equity. High risk. DYOR. Not financial advice. Not available in the US or other restricted territories. tessera.pe/terms