Today is day 70 after the SpaceX ($SPCX) prospectus, and up to 319.0 million shares of Class A common stock become eligible to sell. That is the entire event. It has been on file since June 11, 2026, in a table in the prospectus, along with every other date between now and the middle of 2027.
It is being discussed as though it were not. The framing going round is "the third major unlock," with no number attached — and where a number does appear, it is usually a percentage with the wrong denominator behind it.
This piece is about that denominator, because getting it wrong on this particular stock moves the answer by 2.9x, and because the arithmetic generalises: every staggered lock-up in the last few years works this way, and the same mistake is available in all of them.
The trap: 7% of what
The prospectus releases the day-70 tranche as "7% of the shares subject to the 180-day lock-up period."
Not 7% of the company. Not 7% of shares outstanding. Seven per cent of the locked pool, which is a different and much smaller number than the share count — because it excludes the shares sold in the IPO, excludes Mr. Musk's entire holding, and excludes everything else on a longer lock-up.
Take the two readings side by side:
- 7% of the locked pool = the disclosed 319.0 million shares, which is 2.44% of shares outstanding.
- 7% of shares outstanding = 915.3 million shares.
The wrong reading is 2.87x the right one. And it is an easy mistake to make honestly, because the tranche is quoted as a percentage in the filing and as a share count in press coverage, and the two only reconcile if you know which pool the percentage refers to.
You can recover the pool by division, which is the useful habit here. The first tranche was 20% of the same pool and 911.5 million shares. So the pool is 911.5 ÷ 0.20 = 4.56 billion shares. Check it against the day-70 tranche: 319.0 ÷ 0.07 = 4.557 billion. Check it against the conditional tranche: 455.8 ÷ 0.10 = 4.558 billion. Three disclosures, one pool, no assumptions.
Against 13,075,865,175 shares outstanding after the offering, that pool is about 35% of the company. Everything the next four months release comes out of it.
The schedule, with the denominator supplied
The chart at the top of this page is every share SpaceX has, sorted by whether it can be sold, on each date the answer changes. The table under it is the same schedule as filed, with a column the filing does not carry: what each tranche is as a share of the company.
Three things fall out of it immediately.
The largest tranche of 2026 already happened. August 6, the second trading day after the Q2 results, released up to 911.5 million shares — 2.86x today's tranche, and 6.97% of the company. Whatever supply that produced, the market has already met it.
Today's tranche is one of the smallest on the list. At 319.0 million shares it is smaller than every remaining 2026 date except the day-91 affiliate release.
The one worth a diary entry is after the Q3 print. Up to 1.3 billion shares — 28% of the pool, 9.94% of the company, and 4.08x today. It has no fixed date, because it keys off the Q3 earnings release rather than a day count, which is presumably why it attracts none of the attention that a calendar date does.
By December 8 the entire 180-day pool has released. Add the IPO shares and 40.1% of the company is freely tradable — a figure that matches the "about 40% by December" line circulating without a source, and which turns out to be exactly right once you do the addition.
The condition nobody checked, which made December bigger
There is a row in the prospectus table that did not happen, and it is the most interesting one.
An additional 455.8 million shares — 10% of the pool — would have released on August 6 if the stock had closed at or above 130% of the $135.00 IPO price for at least five of the ten trading days ending on the first earnings release date.
130% of $135.00 is $175.50.
SpaceX closed its first session at $125.33 (as of June 12) and stood at $146.23 at the August 18 close, the price basis our model currently carries. The trigger was never in reach. The condition failed.
That has a consequence the prospectus states and nobody has quoted: the December 8 remainder is "up to 328.4 million shares if the Additional Release Shares were released… or up to 797.6 million shares if the Additional Release Shares were not released." Because the trigger was missed, December 8 is a 797.6 million share release rather than a 328.4 million one — 2.43x larger, and 6.10% of the company.
So the market's own weakness in early August did not avoid a release. It moved 469 million shares from August to December. That is the kind of thing a schedule with conditions in it does, and it is why the conditions are worth reading rather than the headline dates.
The tranche that dwarfs every other
Everything above concerns a pool of 4.56 billion shares. Mr. Musk's holding is 6.4 billion shares on its own — 48.95% of the company — and it is locked for 366 days, releasing on June 12, 2027, with, in the prospectus's words, no early release provisions at all.
Including his stake, shares locked for more than a year total 7.8 billion, which the prospectus puts at approximately 60% of shares outstanding.
That is the shape of this cap table: a 2026 of small, scheduled, pre-announced tranches, and then a single date in June 2027 that is 20x the one being traded today. If a lock-up is going to matter to this stock at all, the calendar says when.
What this does to our model: nothing, and that is the point
Our SpaceX model values the company at a per-share figure built on 13.176 billion shares and a $146.23 price basis as of the August 18 close.
A lock-up expiry changes neither input. It does not issue a share, retire a share, or alter a dollar of revenue, cost or capex. It changes who is allowed to sell, which is a fact about the float and not about the business. There is no line in data/models/spcx.json for it because there should not be one.
This is worth stating plainly because "supply overhang" arguments are usually made in the same breath as valuation ones, and they are not the same kind of claim. A tranche of 319 million shares becoming sellable is a statement about the next few sessions' liquidity. The model is a statement about 2031. The first does not move the second, and anyone telling you an unlock changed what the company is worth is describing a price, not a value.
One reconciliation note while we are here. Our model carries 13.176 billion shares; the prospectus figure used throughout this piece is 13,075,865,175, the post-offering count based on shares outstanding at March 31, 2026. The 100 million share difference is a real one — options and RSU settlements since — and it is ours, not the filing's. Every percentage above uses the prospectus count; on our own share count each one falls by about 0.8% relative, which changes no conclusion here.
What to watch
- Form 4 filings after each tranche. Eligible to sell is not sold. The strongest available evidence of what these dates actually do is whether insiders file anything after them, and so far the interesting fact would be an absence.
- The Q3 earnings date, because it sets the 1.3 billion share release — the only 2026 tranche without a calendar date and the largest one left.
- December 8 at 797.6 million shares, not 328.4 million. Anyone still quoting the smaller figure has not checked whether the August condition triggered.
- The extended lock-up tranches from early 2027 — 351.9 million shares after the Q4 print, 176.0 million on March 18, 2027 — which release a pool of about 1.76 billion held by non-Musk long-term holders.
- June 12, 2027. One date, 6.4 billion shares, no early release, and eleven months of notice.
The lock-up schedule, tranche percentages, share maxima, dates, the $135.00 IPO price, the 130%-of-IPO-price early release condition, the 555,555,555 shares sold in the offering, the post-offering counts of 7,380,196,910 Class A and 5,695,668,265 Class B shares, the two alternative December 8 amounts, Mr. Musk's 6.4 billion shares and 366-day lock-up, and the ~7.8 billion shares (~60% of outstanding) locked for more than a year are all from the "Shares Eligible for Future Sale" and "Lock-Up and Market Standoff Agreements" sections of SpaceX's Rule 424(b)(4) prospectus dated June 11, 2026. The implied 4.56 billion share lock-up pool, the 2.87x wrong-denominator comparison, the $175.50 trigger level, every "% of shares outstanding" figure, the 40.1% freely-tradable-by-December total and the 2.86x, 4.08x, 2.43x and 20x ratios are R40 arithmetic on those disclosures. The $125.33 first close and the $146.23 close of August 18, 2026 are quoted prices carried in our Q2 2026 coverage and in the price basis of our model; this site does not store price history, so they are point-in-time quotes with the dates attached rather than a series anyone can re-derive here, and the statement that the $175.50 condition was not met follows from the August 6 release being reported at the 911.5 million figure rather than at 1.37 billion. Our model's 13.176 billion share count is our own input and differs from the prospectus count by about 100 million shares. Nothing here is a claim about what any holder has sold; every share figure in this piece is a maximum that became eligible.
