The interesting question about analyst disagreement is almost never why are they so far apart. That question assumes the distance is a fact about the company, and it usually gets answered with a paragraph about how hard the company is to value — which is true of every company anyone argues about, and therefore explains nothing.
The prior question is whether the distance is a measurement at all. A price-target range is not observed the way revenue is observed. It is assembled: somebody decides which analysts are in the panel, and — the part nobody publishes — how old a note is allowed to be before it drops out. Change either decision and the range changes without a single analyst changing their mind.
So the question this piece asks about SpaceX is: how much of the famous spread is a disagreement about SpaceX, and how much is a disagreement between data vendors? It is answerable, it takes about ten minutes, and nobody seems to have done it.
Here is what three vendors published for the same stock in the same week.
| Panel | As of | Analysts | Average | Low to high |
|---|---|---|---|---|
| S&P Global, via StockAnalysis | 8 Sep 2026 | 35 | $222.32 | $117 to $450 |
| TipRanks | 8 Sep 2026 | 34 | $231.21 | $75 to $800 |
| MarketBeat | 4 Sep 2026 | 43 | $221.20 | $75 to $800 |
The three averages sit inside $10.01 of each other — 4.5% of their own mean. The three high targets are $450 and $800, a gap of $350, or 78%. Same company, same week. On SpaceX's 13.18 billion shares, that $350 of disagreement about a single analyst's number is about $4.6 trillion of implied value — and it is the statistic most likely to be quoted as "the analyst range."
The centre is a measurement. The tails are a policy
That pattern is not noise, and it is not a coincidence. It falls straight out of how the two statistics are computed.
An average over thirty-odd notes is hard to move. Add or drop two analysts and it barely shifts, which is why three vendors polling overlapping but different sets of people land within $10 of each other. That number is telling you something real about where the sell side is: a little over $220.
A minimum and a maximum are the opposite. Each is a single note, so each is entirely determined by whether that one note is still in the panel. There is no averaging to hide behind. The high is not a summary of anything — it is one analyst, and the question of whether you see it is the question of whether your vendor still counts them.
So when a range widens or narrows, the honest first hypothesis is not that the market's view changed. It is that a note aged out.
The $800 is two months old, and one vendor has stopped counting it
Both halves of the disagreement have names, and the dates are the story.
The $800 is Raymond James. Brian Gesuale initiated coverage at Strong Buy with an $800 target on 9 July 2026, calling Starship "the defining industrial innovation of our generation." TipRanks and MarketBeat both still carry it. The S&P Global panel does not, and its high is instead $450 — Arete Research, which raised from $401 on 11 August. Nothing happened to Raymond James' view in between. One vendor's window is long enough to hold a July initiation and another's is not.
The $62 that circulates as the low is not a price target at all. It is Morningstar's fair value estimate, published 16 June 2026 — a different kind of number, produced by a different process, and present in none of the three panels above. It gets quoted alongside sell-side targets because it is a dollar figure next to a ticker. The lows that are actually in the panels are $75 and $117, and the $75 belongs to an analyst we could not identify from public sources, which is worth saying plainly rather than guessing at.
Two of the three most-quoted numbers in the SpaceX target debate, then, are a two-month-old initiation and a figure from a different discipline.
Our own four numbers are not a distribution, and pretending otherwise would be the same mistake
The temptation here is to put our model beside the panel and note how the extremes line up. It is worth resisting, because the two objects are not the same kind of thing.
A panel range is thirty-odd people each publishing one best guess, and its shape carries information about how much they disagree. Our four scenarios are one team deliberately marking corners. Bear, base, bull and Elon are arguments with names, chosen to bracket, not sampled from anything. The distance between our bear and our Elon case is not a measure of uncertainty. It is a measure of how far apart we decided to put the goalposts.
Which leaves exactly one defensible comparison — centre against centre:
| Value | What it is | |
|---|---|---|
| Street average, three vendors | $221 to $231 | 33–43 analysts, 8 September |
| R40 base case | $187.12 | Our model, as of 19 August |
| R40 bull case | $467.81 | Our model, as of 19 August |
The sell side's centre sits 18% to 24% above our base case, and our bull case is above every live target on the street except the aged $800. Those four scenario values are model output on published assumptions — not forecasts, not targets, and they moved on 19 August when the model gained its Model-layer vertical.
That is a real disagreement, it is about a fifth of the stock, and it is much smaller than the headline range implies. Almost everything about SpaceX is either disclosed or narrow: the June quarter put revenue at $7,814M, up 92% on the year, Connectivity at $4,291M with $1,656M of operating income, and AI solutions and infrastructure at $2,194M. Nobody is far apart on any of that.
The axis all of it lives on is launch rate, and it is not launch price
If the disagreement is concentrated, it is worth knowing on which axis — and in our model it is emphatically not the price of a launch.
Sweep the Space vertical's launch-price drift across its whole plausible band, from −5% a year to zero, and fair value moves from $184.12 to $188.29. The entire range is $4.17 a share. Launch pricing is close to inert.
Cadence is the opposite, and we published the arithmetic on 21 August rather than re-deriving it here: feeding SpaceX's own stated launch ladder into the model in place of ours takes the base case to $500.93, while the one rung that can be graded next August — one flight a day, starting immediately — is worth $7.47. That piece's conclusion was that the entire distance between a $187 stock and a $501 stock is a launch rate. The vendor panels are arguing on the same axis, from the other side. The second chart puts the two levers next to each other.
Pivotal is counting something nobody models, including us
Which brings us to the note that prompted this piece, and to the part where we come off worse than the analyst.
Pivotal Research initiated SpaceX at Buy with a $220 target on 8 September. The thesis rests on Starship reusability requiring "20 to 50 flights per vehicle with relatively inexpensive and quick refurbishment and redeployment" — described as "a single admittedly massive engineering bottleneck," without which SpaceX would be "a different and much smaller company."
Read that quantity carefully, because it is not the one everyone else is arguing about. Flights per vehicle is durability. Flights per year is rate. They are different claims and they are not substitutes. Twenty flights a year flown by fifty vehicles and twenty flights a year flown by one vehicle are identical cadence and completely different businesses — because the second one is the thing that makes a flight cheap.
Our model does not distinguish them. The bull case says Starship "reaches high cadence and low cost." The Elon case says "Starship at a flight a day." Both are rate claims. There is no reflight-count driver in any vertical of our SpaceX model — cost and cadence are assumed to move together, which is a modelling convenience, not a finding. Pivotal at least named the quantity, and as far as we can establish from public write-ups this is the only sell-side SpaceX target built explicitly on flights per vehicle. That is a survey of what is publicly visible, not a claim about what sits in anyone's spreadsheet.
The observed base for the durability claim is thin in a way worth stating: as of our most recent Starship coverage on 29 August, first ship catch and first ship reflight were both still ahead. No Starship upper stage has flown twice. Pivotal's thesis needs twenty to fifty.
Whether our model should carry a durability driver is a real question and this piece does not answer it by adding one. It is now a question about the model, and it belongs on the model's own ledger rather than in an article.
What to watch
- Whether the S&P Global panel's high stays at $450 into October. Raymond James' initiation passes three months on 9 October. If the $450 high survives while TipRanks and MarketBeat still show $800, the divergence is confirmed as a windowing rule rather than a change of view, and the quoted "range" should be read accordingly.
- The first Starship ship reflight, and the turnaround time attached to it. One vehicle flying twice converts Pivotal's thesis from an assertion into a data point, and the interval between the two flights is the number that matters — not the fact of it.
- Whether any second house publishes a target with a flights-per-vehicle figure in it. One note built on durability is an outlier. Two is the beginning of the street pricing a different variable from the one it has been arguing about.
- The next model revision. Our four scenario values last moved on 19 August. A durability driver, if one is warranted, would move the bull and Elon cases specifically, because they are the two that currently assume cost falls with rate.
- The Falcon handoff. If Falcon keeps carrying roughly 150 flights a year while Starship stays in single digits, then the cadence assumptions in every model on this page — ours and the street's — are being underwritten by a vehicle that is not yet doing the work.
The three panels are as published by StockAnalysis.com citing S&P Global Market Intelligence and by TipRanks, both read on 8 September 2026, and by MarketBeat as of 4 September 2026; analyst counts, averages and ranges are theirs. Raymond James' 9 July initiation and Morningstar's 16 June fair value estimate are press-reported. Pivotal Research's 8 September initiation and the quotations from it are press-reported from the note, which is not public. SpaceX's June-quarter revenue and segment figures are as reported, captured in our quarterly notes. The launch-price sweep, the scenario values and the $500.93 and $7.47 cadence figures are output from our own SpaceX model on published assumptions as of 19 August 2026, against a $146.23 basis price that day — model output, not forecasts and not price targets. Percentage comparisons between the panels and our model are our arithmetic.
