SpaceX Marketed the Largest IPO in History on the Weakest Quarter It Had Disclosed
SPCX financial model
Income Statement · Dollars in millions, except per share
| Income Statement | ||||||
|---|---|---|---|---|---|---|
| Actual | Estimate | |||||
| 2023 | 2024 | 2025 | 2026E | 2027E | 2028E | |
| Space Revenue | $3,557.0 | $3,796.0 | $4,086.0 | $3,243.8 | $4,150.1 | $4,648.1 |
| Connectivity Revenue (Starlink) | 3,869.0 | 7,599.0 | 11,387.0 | 15,284.4 | 21,223.0 | 26,727.9 |
| AI Revenue (xAI + X) | 2,961.0 | 2,620.0 | 3,201.0 | 4,027.0 | 6,259.1 | 8,690.4 |
| Total Revenue | $10,387.0 | $14,015.0 | $18,674.0 | $22,555.2 | $31,632.3 | $40,066.4 |
| Less: Cost of Revenue | ($6,110.0) | ($7,996.0) | ($9,451.0) | ($11,223.2) | ($15,089.6) | ($18,230.2) |
| Gross Profit | $4,277.0 | $6,019.0 | $9,223.0 | $11,331.9 | $16,542.7 | $21,836.2 |
| Less: Research & Development | ($2,105.0) | ($3,464.0) | ($8,643.0) | ($13,825.2) | ($14,391.5) | ($14,423.9) |
| Less: Selling, General & Administrative | (1,665.0) | (1,813.0) | (2,644.0) | (3,419.1) | (4,301.9) | (5,008.3) |
| Less: Restructuring Charges (Credits) | (237.0) | (213.0) | (487.0) | 11.0 | 0.0 | 0.0 |
| Less: Impairment | (3,775.0) | (63.0) | (38.0) | 0.0 | 0.0 | 0.0 |
| Total Costs and Expenses | ($13,892.0) | ($13,549.0) | ($21,263.0) | ($28,456.5) | ($33,782.9) | ($37,662.4) |
| Income (Loss) from Operations | ($3,505.0) | $466.0 | ($2,589.0) | ($5,901.4) | ($2,150.7) | $2,404.0 |
| EBITDA | ($870.0) | $4,290.0 | $4,112.0 | $6,611.7 | $15,967.7 | $21,423.1 |
| Interest Income | $249.0 | $371.0 | $492.0 | $2,531.6 | $3,737.6 | $3,478.0 |
| Interest Expense | (1,693.0) | (1,580.0) | (1,945.0) | (2,247.5) | (2,197.6) | (2,184.2) |
| Other Income (Expense), Net | (42.0) | 985.0 | (177.0) | (1,876.0) | 0.0 | 0.0 |
| Income (Loss) Before Income Taxes | ($4,991.0) | $242.0 | ($4,219.0) | ($7,493.2) | ($610.6) | $3,697.8 |
| Less: Provision for (Benefit from) Income Taxes | $363.0 | $549.0 | ($718.0) | ($54.3) | ($9.2) | ($554.7) |
| Net Income (Loss) | ($4,628.0) | $791.0 | ($4,937.0) | ($7,547.6) | ($619.8) | $3,143.1 |
| Ratios & Assumptions | ||||||
| YoY Total Revenue Growth | 34.9% | 33.2% | 20.8% | 40.2% | 26.7% | |
| YoY Space Revenue Growth | 6.7% | 7.6% | (20.6%) | 27.9% | 12.0% | |
| YoY Connectivity Revenue Growth | 96.4% | 49.8% | 34.2% | 38.9% | 25.9% | |
| YoY AI Revenue Growth | (11.5%) | 22.2% | 25.8% | 55.4% | 38.8% | |
| Gross Margin (memo) | 41.2% | 42.9% | 49.4% | 50.2% | 52.3% | 54.5% |
The full workbook adds 2 historical and 8 projected quarters, plus KPI Drivers · Balance Sheet · Cash Flow Statement — every subtotal a live formula, every projection traced to a driver.
Key Takeaways
- This is the quarter the prospectus was built on, and it was the weakest of the four SpaceX has disclosed at quarterly granularity. Revenue grew 15.4% to $4,694M, the slowest growth in the disclosed record. Operating income went from a $27M profit a year earlier to a $1,943M loss. Adjusted EBITDA fell 34.9%, and net loss widened roughly eightfold to $4,276M.
- Connectivity was the only segment with an operating profit, and its margin was shrinking. The segment grew 31.6% to $3,257M and earned $1,188M at the operating line, but its margin fell 526 basis points to 36.5%. The pressure showed up in costs: selling, general and administrative expense doubled, while consumer revenue still rose 44% despite a 22.9% fall in ARPU.
- The AI segment spent $7,723M of capital to produce $818M of revenue and a $2,469M operating loss. Research and development alone was $2,379M, up 162% and nearly three times segment revenue. The segment's growth that quarter came from Grok and X subscriptions, not from compute. The cloud contracts behind its later turn to positive adjusted EBITDA show up in the following quarter.
- Capex reached 215% of revenue, and one debt-extinguishment charge explains most of the gap to net loss. The company spent $10,107M against $4,694M of sales, for free cash flow of about $(9.1)B. Below the operating line, a $1,526M loss on extinguishment of debt led $1,876M of other expense, and with $451M of net interest took the pre-tax loss to $4,270M.
- Rating: maintaining Outperform, and this retrospective carries no rating action. The standing call was set on the Q2 print and nothing here changes it. What this quarter does change is the reading of that print: the Q2 inflection is steeper and more recent than a reader starting from the Q2 report alone would understand.
Why This Quarter Never Got a Report
SpaceX filed its S-1 on May 20, 2026. The most recent completed quarter in that document was Q1 2026, which had closed seven weeks earlier. The IPO priced at $135 on June 11 and the shares began trading on June 12. The first earnings release the company ever issued as a public company was the Q2 print on August 4.
The consequence is a gap that is easy to miss. Q1 2026 is the quarter investors actually bought, in the sense that it was the last full period of financial history available when they made the decision. It is also the last quarter SpaceX reported without an earnings release or a call. It appeared in the prospectus as one column among many, and it reappeared three months later as a comparative column in a release whose headline was about a different quarter.
Read in isolation, Q1 2026 is an uncomfortable document. Read against what came next, it is the most useful quarter in the company's short public record, because it establishes the base from which the Q2 inflection was measured. A reader who knows only the Q2 report knows that revenue grew 92% year over year and that operating loss narrowed to $143M. That reader does not know that the sequential comparison was against a quarter in which the company lost $1,943M at the operating line and spent more than two dollars of capital for every dollar of revenue.
Results
There is no consensus column in the table below, and there cannot be one. The comparison that matters is year over year, and the prospectus prints Q1 2025 beside Q1 2026 for exactly that purpose.
Q1 2026 versus Q1 2025
| Metric | Q1 2026 | Q1 2025 | Change | Read |
|---|---|---|---|---|
| Revenue | $4,694M | $4,067M | +15.4% | Slowest growth in the disclosed record |
| Space revenue | $619M | $865M | −28.4% | Fewer customer launches |
| Connectivity revenue | $3,257M | $2,475M | +31.6% | The only segment growing and profitable |
| AI revenue | $818M | $727M | +12.5% | Subscriptions up, advertising down |
| Income (loss) from operations | $(1,943)M | $27M | swing of $(1,970)M | From breakeven to a 41.4% operating loss margin |
| Adjusted EBITDA | $1,127M | $1,730M | −34.9% | Margin 24.0% against 42.5% |
| Net loss | $(4,276)M | $(528)M | 8.1x wider | Operating swing plus a $1,526M debt-extinguishment loss |
| Capex | $10,107M | $4,140M | +144.1% | 215.3% of revenue against 101.8% |
Q1 2025 figures are as printed in the S-1 and the final prospectus.
Q1 2026 against the quarter that followed
| Metric | Q1 2026 | Q2 2026 | Change |
|---|---|---|---|
| Revenue | $4,694M | $7,814M | +66.5% |
| Income (loss) from operations | $(1,943)M | $(143)M | +$1,800M |
| Adjusted EBITDA | $1,127M | $3,538M | +213.9% |
| Net loss | $(4,276)M | $(541)M | +$3,735M |
| Capex | $10,107M | $18,369M | +81.7% |
| Starlink subscribers | 10.3M | 12.0M | +1.7M |
| Nameplate compute | 1.0 GW | 1.4 GW | +0.4 GW |
Segment Performance
| Segment | Revenue | YoY | Operating income | Operating margin | Capex | Capex / revenue |
|---|---|---|---|---|---|---|
| Space | $619M | −28.4% | $(662)M | −106.9% | $1,052M | 170% |
| Connectivity | $3,257M | +31.6% | $1,188M | 36.5% | $1,332M | 41% |
| AI | $818M | +12.5% | $(2,469)M | −301.8% | $7,723M | 944% |
| Total | $4,694M | +15.4% | $(1,943)M | −41.4% | $10,107M | 215% |
Space: fewer customers, more constellation
Space revenue fell 28.4% to $619M while the segment put 23.6% more mass into orbit than a year earlier. Those two facts are connected. The company attributes the revenue decline to fewer customer launches: launch services revenue fell $236M, or 41.7%, to $330M, and launch and development revenue slipped $10M on the timing of government contract work. Customer launches fell from 12 to 7 and customer payloads from 75 tonnes to 45, while internal payloads rose from 375 tonnes to 511. Falcon launches rose from 36 to 40, and a larger share of that capacity went to deploying Starlink.
Costs moved the other way. Research and development rose $404M, or 76.8%, to $930M, which is 150% of segment revenue. Management attributes the increase to accelerated Starship development and to the production and launch facilities that will support it, ahead of the first V3 suborbital flight in May.
Assessment: the revenue loss reads as capacity going to the constellation rather than demand disappearing, and the Q2 recovery to $962M of revenue on a richer customer mix supports that reading. The cost is that Space stopped covering its own spending: a segment that lost $70M at the operating line in Q1 2025 lost $662M in Q1 2026.
Connectivity: growing fast, earning less on each dollar
Connectivity grew 31.6% to $3,257M and produced $1,188M of operating income, the only operating profit in the company. Consumer revenue rose 44.0% to $2,148M: subscribers grew 104.7% to 10.3 million while ARPU fell 22.9%, from $86 a month to $66. Enterprise and government revenue grew only 12.8% to $1,109M. Growth of $209M in aviation, maritime and other enterprise business and $85M in mobile connectivity was partly offset by a $175M decline in government connectivity, and the line's share of the segment fell from 39.7% to 34.1%.
The margin story is the one that matters. Operating margin fell 526 basis points to 36.5%, because segment costs grew 43.5% against revenue growth of 31.6%. Measured against revenue, selling, general and administrative expense, which more than doubled (+102.9%), accounts for 230 of those points; research and development, up 66.7%, for 132; and cost of revenue, up 36.0%, for 164. Lower ARPU did not stop consumer revenue rising 44%, but it is part of why revenue grew more slowly than costs. Management said in the prospectus that it expects ARPU to keep declining over the next few years as the base outside North America grows, as lower-priced plans are added and as it adjusts broadband plan fees, and that it expects scale and technological advancement to offset the effect.
Assessment: Q1 was a weak quarter for this segment but not its weakest disclosed one. Connectivity's operating margin was lower in Q2 2025, at 35.7%, and so was its enterprise share, at 33.5%. What Q1 adds is where the compression showed up: mostly in operating expense, the more controllable part of the cost base. The enterprise share then rose to 42.1% in Q2.
AI: nine hundred percent of revenue spent on capital
The AI segment produced $818M of revenue and a $2,469M operating loss, against $7,723M of capital expenditure. Research and development of $2,379M was 162% higher than a year earlier and nearly three times the segment's revenue. The segment exited the quarter with 1.0 gigawatt of nameplate compute, up from 0.3 a year earlier.
Beneath the total, the two revenue lines moved in opposite directions. Advertising revenue fell $100M, or 22.6%, to $343M, which management attributes to an overhaul of the advertising platform that disrupted ad sales for a short period during the rebuild. AI solutions and infrastructure revenue rose $191M, or 67.3%, to $475M, and crossed above advertising for the first time in the disclosed record. But $177M of that increase came from Grok and X subscriptions and $12M from data licensing. In Q1 the line grew on subscriptions. The cloud services agreements that added $1.6B of infrastructure revenue in Q2 are the reason the segment looks different today.
Assessment: at Q1 the AI segment consumed 76% of consolidated capex to generate 17% of consolidated revenue at a negative 302% operating margin. One quarter later it produced $2,561M of revenue and positive adjusted EBITDA of $1,146M, on $14.1B of newly contracted cloud services agreements. The Q1 revenue mix is why that result should be read as a step rather than a slope: the capacity was built in Q1, and the contracts that monetised it produced their revenue afterwards.
Key Operating Metrics
| KPI | Q1 2026 | Q1 2025 | Change |
|---|---|---|---|
| Total launches | 40 | 38 | +5.3% |
| Customer launches | 7 | 12 | −41.7% |
| Internal launches | 33 | 26 | +26.9% |
| Mass to orbit | 556 t | 450 t | +23.6% |
| Customer payloads | 45 t | 75 t | −40.0% |
| Internal payloads | 511 t | 375 t | +36.3% |
| Starlink subscribers | 10.3M | 5.0M | More than doubled |
| Starlink ARPU | $66/mo | $86/mo | −23% |
| Nameplate compute | 1.0 GW | 0.3 GW | +0.7 GW |
Subscriber, ARPU, compute, total-launch, customer-launch and mass-to-orbit figures are as printed in the prospectus, which puts subscriber growth at 104.7% and the ARPU decline at 22.9%. The prospectus counts 36 and 40 Falcon launches in the two quarters. The other two launches in Q1 2025 were not Falcon flights, and the prospectus classifies all Starship launches as internal. The Q1 2025 payload split is derived from the Q2 8-K's first-half and second-quarter columns.
The Capital Cycle
Capital expenditure of $10,107M against revenue of $4,694M is the single number that separates this quarter from anything a conventional software or telecom comparison would suggest. The ratio was 215%. A year earlier it was 102%, which is itself extraordinary. Three quarters of the Q1 2026 spend, $7,723M, went into the AI segment.
Operating cash flow of $1,047M against that capex left free cash flow of about $(9.1)B for the quarter. At March 31 the company held $15.9B of cash and equivalents and $7.8B of short-term securities, against $29.1B of debt principal, with $1.5B available under its credit facility. At the Q1 rate, cash alone covered about 1.8 more quarters, and cash plus securities about 2.6. On that arithmetic the $86B raise, including the fully exercised over-allotment, looks closer to necessary than opportunistic.
Assessment: the capital intensity is the bear case in its purest form, and Q1 shows it plainly, because it is the quarter where the spending had already arrived and the revenue had not. The Q2 print does not resolve the question. It raises the spend to $18,369M, and the ratio to 235%, while finally showing what the earlier capital bought.
Below the Operating Line
The operating loss was $1,943M. The net loss was $4,276M. Most of the $2,333M difference is a single charge.
| Item | Q1 2026 | Note |
|---|---|---|
| Loss from operations | $(1,943)M | As printed |
| Interest expense | $(664)M | On $29.1B of debt principal at quarter end |
| Interest income | $213M | On $23.7B of cash and short-term securities at quarter end |
| Other income (expense), net | $(1,876)M | The largest below-the-line item |
| of which: loss on extinguishment of debt | $(1,526)M | One-time |
| of which: unrealized loss on digital assets | $(344)M | Mark-to-market |
| Loss before income taxes | $(4,270)M | As printed |
| Provision for income taxes | $(6)M | Negligible |
| Net loss | $(4,276)M | As printed |
| Net loss attributable to common shareholders | $(4,947)M | $671M wider than net loss |
Two observations. First, other expense of $1,876M equals 40% of consolidated revenue and is more than twice the AI segment's revenue for the period. The prospectus attributes it mainly to the loss on extinguishment of debt and the unrealized loss on digital assets, which together account for $1,870M of it. The Q2 figure for the same line was $86M, so this was a Q1 event rather than a run rate.
Second, net loss attributable to common shareholders was $671M wider than net loss in the first half, and all of that gap falls in Q1. In Q2 the two figures are identical at $541M.
Assessment: excluding the $1,526M extinguishment charge, the Q1 net loss would have been about $2.75B rather than $4.28B. The MD&A names the charge, but its $1,526M amount appears only in the cash-flow statement and the debt note; the income statement shows a single other-expense line. A reader working from the headline figures alone would take the quarter to be worse than its recurring economics were. On this line the quarter's weakness is overstated, and that cuts in the company's favour.
What the Prospectus Said About It
Every weakness above is in the prospectus, with management's explanation beside it. The quarter had no release and no call, but it did have an MD&A. Its explanations, in brief:
- Other expense: primarily the loss on extinguishment of debt and the unrealized loss on digital assets.
- ARPU: expected to keep declining over the next few years as the subscriber base outside North America grows, lower-priced plans are added and broadband plan fees are adjusted, with scale and technological advancement expected to offset the effect.
- Advertising: down because an overhaul of the advertising platform disrupted ad sales for a short period during the rebuild.
- AI solutions and infrastructure growth: primarily Grok and X subscription revenue, plus data licensing.
- Space revenue: fewer customer launches, and the timing of work on government contracts.
- Space research and development: accelerated Starship development and the facilities to support future Starship launches.
Our read: the disclosure was complete. What the quarter lacked was a release and a call to put these explanations in one place. Taken together, they describe a quarter that looked worse than its recurring economics on the net loss line, and roughly as bad as it looked at the operating line.
Market Context
This quarter has no market reaction, and no honest version of this note can supply one. The shares did not trade during Q1 2026 and did not begin trading until June 12, three weeks after the S-1 made the quarter public. The IPO priced at $135 on June 11, against a pre-pricing indication of about $1.75 trillion, and closed on June 15 with 638,888,888 shares sold, including the full over-allotment, for gross proceeds of about $86.25B.
What can be said is that these were the numbers in front of every investor who participated in that pricing. The first market judgement on the combined entity was rendered on a quarter with 15% revenue growth, a 41% operating loss margin, and capital expenditure at more than twice revenue. Whatever one concludes about the valuation achieved, it was not achieved on the strength of the most recent quarter's operating performance.
The Debates This Quarter Frames
There was no sell-side debate at the time, because there was no sell-side. The debates below are the ones the quarter sets up, and the ones the subsequent record has begun to answer.
Debate: was Q1 the trough or the trend?
Bull view: Q1 is the bottom of a deliberate investment cycle. Space gave capacity to the constellation, AI spent ahead of contracted revenue, and both showed returns in the following quarter. Revenue grew 66.5% sequentially and operating loss narrowed by $1,800M.
Bear view: one quarter of recovery on newly signed cloud agreements is not proof of a cycle turning. The Q1 cost base did not shrink in Q2, it grew. The improvement came entirely from revenue arriving, and that revenue rests on contracts that can terminate on 90 days notice after ramp.
Our take: the bulls have the better of this, but by less than the Q2 headline suggests. The honest reading is that Q1 shows how much of the Q2 result was a step function rather than a trajectory, and a step function needs a second confirming quarter before it becomes a trend.
Debate: is Connectivity margin compression structural?
Bull view: ARPU decline is the signature of a consumer network scaling into new geographies. The 526 basis points of margin given up in Q1 came with a subscriber base that more than doubled year over year and reached 12 million by June, and enterprise and government revenue grew 63% sequentially in the following quarter.
Bear view: cost growth outran revenue, ARPU fell 23%, and the enterprise share fell in Q1. The company itself expects ARPU to keep falling. A network business that keeps cutting price to add subscribers has not demonstrated pricing power.
Our take: Q1 still produced a 36.5% operating margin, inside the 35.7% to 41.7% range of the four disclosed quarters, and most of the compression showed up in operating expense. That is an argument for the pillar, not against it.
Debate: does the capital intensity have a ceiling?
Bull view: capital spending tracks contracted demand. The Q1 build produced the compute that generated $2,561M of AI revenue in Q2, and the $100B cash and securities position funds the programme.
Bear view: capex rose from $10.1B to $18.4B in one quarter with guidance for two more at that level, which is a run rate no disclosed revenue path covers. Q1 shows how far spending can run ahead of revenue, and nothing establishes that the lag stays short.
Our take: this remains the open question in the thesis and Q1 sharpens rather than settles it. The quarter demonstrates both that the capital works and that it works on a delay long enough to be dangerous if demand pauses.
Model Implications
This is a historical quarter and it changes no forward estimate. The model already carries Q1 2026 and Q1 2025 as printed in the prospectus. When it is rolled forward to Q2 actuals, three points apply.
| Item | Recommendation | Reason |
|---|---|---|
| IPO proceeds | Book about $85.7B of net proceeds (638,888,888 shares, $86.25B gross), not $74.4B | The model records the 555,555,555-share base deal. The underwriters' option was exercised in full at closing. |
| Connectivity margin path | Anchor on the 35.7% to 41.7% range of the four disclosed quarters, not on Q2 alone | One quarter is not a trend, and Q1 shows the margin moves with operating expense growth |
| AI segment revenue | Drive it from contracted cloud services, not from the Q1 run rate | Q1 growth was subscriptions. The 213% sequential step came from new contracts and should not be extrapolated. |
Valuation impact: none. The fair value range of $1.65 trillion to $1.95 trillion set on the Q2 print stands, and a retrospective on a prior quarter is not a basis for moving it.
Thesis Scorecard, Read Backwards
The standing thesis was set on the Q2 print. The table below grades each pillar against what Q1 shows, which is a test of whether the pillars hold at the weakest point in the disclosed record rather than the strongest.
| Thesis point | Status at Q1 | Notes |
|---|---|---|
| Bull #1: Connectivity is the economic engine | Confirmed | $1,188M of operating income against a consolidated loss of $1,943M. The engine was carrying the company even as its own margin compressed 526 bp. |
| Bull #2: Launch services is a contracted cash generator | Challenged | Launch services revenue fell 41.7% on fewer customer launches and the segment lost $662M. The cash generation was not visible in this quarter at all. Consistent with the AT RISK tag the standing thesis already carries. |
| Bull #3: Starship is an option on full reusability | Neutral | $930M of Space research and development, up 76.8% on accelerated Starship work, ahead of the May V3 flight. Cost visible, result not yet. |
| Bull #4: AI is a scaling infrastructure business | Challenged | At Q1 this pillar had no support: $818M of revenue, a $2,469M operating loss, negative adjusted EBITDA, and growth that came from subscriptions rather than compute. The pillar is a Q2 result. |
| Bear #1: Valuation demands underwriting years of compounding | Confirmed | The IPO was marketed against a pre-pricing indication of about $1.75T on a quarter annualising to under $19B of revenue with a 41% operating loss margin. |
| Bear #4: Customer concentration and 90-day contract terms | Neutral | The prospectus discloses concentration for full years only: one customer was 20.9% of 2025 revenue, against 24.2% in 2024. The two-customer exposure the standing thesis tracks is a Q2 development. |
| Bear #6: Capital intensity | Confirmed | 215% of revenue against 102% a year earlier. Q2 went higher still at 235%, so Q1 is where the ratio broke rather than where it peaked. |
Overall: the standing thesis survives its weakest quarter with one bull pillar confirmed, one neutral and two challenged, and two bear points confirmed. That is a materially different picture from the Q2 scorecard, and it is the correct picture of the risk an IPO participant took.
Action: no change. Outperform stands on the Q2 evidence. The value of this quarter is calibration rather than decision: it sets the base against which the next two prints should be judged, and it establishes that the Q2 inflection was a step rather than a slope.