SpaceX's satcom pricing is not cost-anchored and cannot be beaten on price by any competitor modeling Starlink as a standalone business.
The Q2 2026 10-Q (primary, sources/latest/edgar-docs/) established the cross-subsidy at segment grain: Connectivity is the only profitable segment (revenue $4.29B, operating income $1.66B), funding a Space operating loss of $542M and an AI operating loss of $1.26B — though AI's Segment Adjusted EBITDA also turned positive for the first time (+$1.15B), on Colossus compute-rental revenue. AI capex was $15.83B of an $18.37B group total, and CFO guidance confirms that run-rate holds through at least Q4 2026 — against Street H2 revenue consensus, roughly 1.6x H2 top line (tracked series). Three capital-intensive commitments layer on top: the $60B Cursor/Anysphere acquisition (CLOSED Aug 14, VWAP-priced), the $16.8B-first-phase/up-to-$119B Terafab chip complex with Tesla, and an ownership/related-party concentration — director Antonio Gracias's Valor entities hold $13.3B of AI-lease debt and 6.5% of Class A stock, now joined by passive Founders Fund (5.5%) and Google (7.2%) filings (tracked series) — a governance watch item, not yet a red flag.
(a) SpaceX raising *new-subscriber* list prices in markets with no live LEO competitor while guiding ARPU upward group-wide
ARPU has now held flat at $66 across three consecutive reads (tracked series), with no upward group guide
(b) SpaceX unable to complete a *future primary* raise on acceptable terms, or a disorderly capital-market event
and now on its strongest evidentiary footing yet: the equity absorbed the Aug 6 lockup, closed above the $135 IPO price on 6 of the 8 trading days from Aug 10-19, and the Cursor merger closed on its original announced timeline without renegotiating the headline figure (tracked series). As long as Musk retains capital-markets access, below-cost Starlink pricing in contested markets is not self-limiting
NVIDIA holds $21.0B (its converted xAI stake) and Saudi PIF holds $26.3B (a fresh, first-ever position, 69.5% of its disclosed US book); AMD and Harvard hold smaller stakes (tracked series). A Colossus chip supplier is simultaneously a top-3 equityholder — detailed in structural dynamics.
The Aug 20 second lockup tranche (~319M shares) coincided with a dip back below the $135 IPO price, unlike Aug 6's same-day +6.14%, though the stock recovered to $141.10 by Aug 26 (tracked series). The $100B Starbase Louisiana megasite and the Cape Canaveral Falcon 9-to-Starship handoff extend the capex story past AI into launch — cadence mechanics owned by launch-bottleneck.
Starlink's June $10/mo equipment fee was dropped by July, and the femtocell plan drew a second skeptical read (Farrar: "something of a bluff"), balanced by Bernstein's new $248 bull case (structural dynamics).
The Aug 24 Colossus 2 injunction hearing did not occur — postponed by an Aug 21 court notice, no new date found (political & regulatory).
Every dollar of AI-segment loss funded by equity and debt still extends SpaceX's capacity to price satcom below cost, and the 13F ownership wave sharpens rather than weakens that read: when a GPU supplier (NVIDIA), a sovereign fund (PIF) and an endowment (Harvard) are all also equityholders, capital access is diversified across counterparty types, not concentrated in one fragile source. "Starlink pricing aggression CONFIRMED INDEFINITE" remains the strongest-evidenced base rate this dossier carries, underwriting "SpaceX financial architecture / Starlink pricing aggression" directly and as a standing assumption in "2 GHz MSS spectrum allocation" and "O3b mPOWER sovereign pipeline."
Two nuances temper, without reversing, that read: Farrar's terminal-capacity argument means capital-unconstrained pricing still runs on production-bounded *broadband* volume (~2M net adds/quarter) — D2D (Africa launch, 14-market target) is a volume lever outside that ceiling and a new, direct touch on SES's African GEO base; and the Poland roaming reversal shows a single government's own service-payment leverage can force a same-week Starlink policy change, a narrower tool than spectrum authority that O3b mPOWER/IRIS² counterparties may reach for sooner.
Starbase Louisiana and the Falcon 9-to-Starship Florida handoff (structural dynamics) reinforce the launch-supply constraint launch-bottleneck tracks against SES's own meoSphere/O3b mPOWER manifest — not restated here. The AI1/Colossus distinction still matters for SES's "space-data / multi-mission platform" adjacency thesis (priors.md OPPORTUNISTIC): Colossus-style terrestrial brokering remains the near-term threat model, not orbital compute, though NVIDIA's Starcloud backing raises the credibility of that category's capitalization.
A new blind spot surfaced this pass, the same species as the orbital-compute undercounting this dossier's own SUPERSEDED history logged three times running: environmental/congressional-oversight developments (Mississippi's Jul 30 turbine agreed order, Pallone's House E&C letter) sat unswept for ~3 weeks because this desk has no standing sweep of state environmental dockets or congressional-committee letters, only FCC/EDGAR. The lesson generalizes past orbital compute: any regulatory/political thread this dossier tracks that lives outside FCC/SEC needs its own periodic direct check, not just a wait for trade-press pickup.
The SPCX daily-close divergence this dossier has carried as "provisional" since Aug 7 is now resolved, not just managed: two independent sources (stockanalysis.com, marketbeat.com) agreed to within $0.12 across the full Aug 10-19 range this pass, cross-checked live — worth noting as the provisional-pricing caution's normal resolution path, not a standing open problem.
Farrar's Apr 2026 "flattering" critique (Structural Dynamics), engaged this pass for the first time, is a reminder that this dossier's disconfirming- evidence hunt has skewed toward the *most recent* Farrar/analyst posts — useful but incomplete, since a genuinely load-bearing critique of the core segment-profitability claim sat unengaged for four months. Worth a standing practice: periodically re-scan a landscape source's back-catalogue for arguments never folded in, not only its newest post, the same discipline already applied to Farrar's Feb 8 Starship piece on 08-07.
A conflicting-dollar-figure problem resolved cleanly this pass by going to the primary filer instead of picking a side: the Aug 20 brief flagged two outlets disagreeing on Saudi PIF's SpaceX stake ($21.5B vs. $26.38B); rather than adopt either, this pass fetched both NVIDIA's and PIF's own 13F-HR `informationTable.xml` directly from EDGAR, which settled it ($26.34B, the filing's own fair-value mark) and supplied NVIDIA's exact share count no secondary source carried. Same lesson as Verify-before-citing (16)/(29)/(35) applied to a *third-party* filer's numbers, not just this desk's own claims or a tracked entity's: a 13F/13G is as fetchable as an 8-K and should be the first stop, not the last, when secondary dollar figures disagree.
A second-source lockup-date correction is a reminder not to over-trust a single secondary aggregator even after it's been right before: purepowerpicks/MLQ's tranche-date schedule (carried since 08-19) proved wrong on the very next tranche's actual date (Aug 20 vs. the tracked Aug 31) once that date arrived and could be checked against live reporting — a "secondary-aggregation, not read off the lockup agreement itself" caveat this dossier already carried on that row, vindicated rather than surprising, but worth remembering the caveat was earned, not decorative.
Will AI1 sign its first commercial contract before/after its ~2027 launch, at what kW?
Resolves when8-K/10-Q or named reporting; on-orbit performance.
Does SpaceX build a Vandenberg pad or resolve the SSO/inclination mismatch?
Resolves whenFAA filing or permitting docket (cross-ref launch-bottleneck).
Does DoD-concentration scrutiny produce a binding multi-vendor mandate, despite the NSSL ceiling tripling the month Reed's criticism surfaced?
Resolves whenNDAA language, GAO report, award pattern.
Do Google's/Reflection's Colossus contracts survive their cancellation checkpoints (Google: Sep 30), and does the Pentagon capacity deal get signed?
Resolves when8-K amendment, non-renewal, or signed contract.
Does bond-spread widening stabilize, and does SpaceX attempt a *new* primary raise?
Resolves whenspread trend; new SEC offering filing.
Does the ~$18.4B/quarter capex rate — holding through Q4, not moderating — accelerate further, and how is negative FCF funded?
Resolves whenQ3 10-Q capex/cash-flow; Cursor's/Terafab's balance-sheet treatment.
Does Flight 14 fly orbital, inserting V3 satellites operationally?
Resolves whenflight outcome, catalogued orbital elements (planet4589); scheduling owned by launch-bottleneck.
Now that the Aug 24, 2026 hearing was postponed with no new date, when does it actually occur, and does the ruling turn on the Jul 30 agreed order's removal-to-Jul-2027 timeline as the "mobile source" question?
Resolves whena new hearing date; the ruling.
Does any orbital-compute actor in either bloc sign a named anchor tenant?
Resolves whena contract or funding round naming a customer.
What is Tesla's financial role in Terafab, and how is the $16.8B first phase (to $119B total) split across SpaceX's/Tesla's balance sheets?
Resolves whenan 8-K/10-Q or JV filing.
Does "Customer B" (AI-only, ~60% of AI Solutions & Infrastructure revenue) resolve to a named counterparty, persisting or diversifying in Q3?
Resolves whenQ3 10-Q footnote.
Does Musk's Aug 11 claim — AI revenue to exceed Connectivity+Space combined by Sept/Q4 — materialize in Q3, or does Connectivity's operating income remain the real funder?
Resolves whenQ3 10-Q segment statement.
Does the widening equity-cross-holding pattern — Gracias/Valor's director + lease-financier + 6.5% role, NVIDIA's supplier + $21B-equityholder role, Founders Fund/Google/PIF/Harvard's passive stakes — draw governance or conflict-of-interest scrutiny from any angle (proxy, press, regulator)?
Resolves whena proxy statement or governance-focused reporting.
Does Farrar's terminal ceiling (~2M net adds/quarter) hold as SpaceX's binding constraint, or does the company scale faster?
Resolves whena production disclosure or net-add print above 2M/quarter.
Does SpaceXAI ever respond to Pallone's Aug 11 document-production demand, and does House E&C escalate (subpoena, further hearing) if not?
Resolves whena committee statement or a SpaceX/xAI response.
Does MoffettNathanson's MVNO-leverage read on the femtocell plan prove out — does AT&T, Verizon, or T-Mobile sign a deal, or does SpaceX pursue standalone buildout regardless?
Resolves whena carrier-partnership announcement or continued standalone capex disclosure.
Complete append-only series — 163 readings, 14 May 2025 → 26 Aug 2026. Rows never change once added; the time series is the evidence base. Each row number is a permanent link — click it to copy a link straight to that reading.
| # | Date | Metric | Value | Source |
|---|---|---|---|---|
| 1 | 2025-FY | Starlink revenue | $11.4B (+48% YoY) | SPCX S-1, brief 2026-07-01 |
| 2 | 2025-FY | Starlink operating profit | $4.4B | SPCX S-1, brief 2026-07-01 |
| 3 | 2025-FY | xAI operating loss | $6.35B | SPCX S-1, brief 2026-07-01 |
| 4 | 2025-FY | Group GAAP net loss | $4.9B | SPCX S-1, brief 2026-07-01 |
| 5 | 2025-FY | Group adjusted EBITDA | $6.6B | SPCX S-1, brief 2026-07-01 |
| 6 | 2026-06-12 | IPO price / Day-1 close | $135 / $161 (+19%) | brief 2026-07-01 |
| 7 | 2026-06-12 | Implied valuation | >$2T | brief 2026-07-01 |
| 8 | 2026-06-08 | AI1 platform spec | 70m, ~150kW peak/120kW sustained, liquid radiator | brief 2026-06-25 |
| 9 | 2026-01-30 | AI1 FCC filing scale | up to 1,000,000 sat data centers | brief 2026-06-25 |
| 10 | 2026-06-08 | AI1 orbital satellite — commercial contract status | ZERO signed; Anthropic has "expressed interest" only; 2 prototypes NET early 2027 | WebSearch, named reporting (bit.com, DCD, techtimes.com, Jun 2026) — corrects 2026-07-05 dossier text, see Superseded |
| 11 | 2026-01 (~) | xAI Series E raise | $20B at $230B valuation (upsized from $15B target) | WebSearch aggregation, named reporting |
| 12 | 2026-02 | SpaceX-xAI merger implied combined valuation | ~$1.25T | WebSearch aggregation, named reporting |
| 13 | 2026-03 | SpaceX $20B bridge loan (Goldman-led) refinances $17.5B X/xAI debt | rate 12.5%→4.58%; combined interest cost ~$900M/yr | WebSearch (octus.com, thenextweb, named reporting) |
| 14 | 2026-05 (mid) | Starlink US/CA/UK price increase | $5-10/mo (US Res100 $50→$55, Res200 $80→$85, Max $120→$130; Standby $5→$10); new subs immediate, existing June 18 | WebSearch (PCMag, 5gstore, driveteslacanada, thenextweb — cross-corroborated) |
| 15 | 2026-06-04 | Starlink subscriber milestone | 12M, 111 days after 10M (Feb) — slower than 52-day (9M→10M) and 48-day (8M→9M) prior intervals | tmfassociates.com, "BARKing mad" (Farrar, Jun 4), citing @Starlink |
| 16 | 2026-06-05 | Google-SpaceX Colossus compute deal | $920M/mo, Oct 2026-Jun 2029 (ramps from Sep 2026 milestone), ~110K GPUs, ~$30B lifetime; 90-day cancellation after Dec 31 2026; terrestrial (Colossus), not orbital | TechCrunch, CNBC (Jun 5, named reporting) |
| 17 | 2026-06-22 | Reflection AI-SpaceX Colossus compute deal | $150M/mo from Jul 1 2026, up to $6.3B through 2029; 90-day cancellation after first 3 months; terrestrial | CNBC (Jun 22), TechCrunch |
| 18 | 2026-06-26 | SpaceX $25B multi-tranche bond offering | closed; ~$90B orders (~3.6x oversubscribed); Baa1/BBB/BBB investment-grade; repays bridge loan | WebSearch aggregation, named reporting |
| 19 | 2026-Q1 | Starlink ARPU | $66/mo (down from $86 Q1 2025, $99 in 2023) | SPCX S-1 (via secondary aggregation — Morningstar, valueaddvc, muskonomy; direct EDGAR fetch 403-blocked) |
| 20 | 2026-Q1 | Starlink net adds | 1.4M (vs ~1.6M Q4 2025) | tmfassociates.com "BARKing mad" (Jun 4), citing S-1 |
| 21 | 2026-Q1 | Starlink operating income | $1.19B (vs $1.03B Q1 2025) | SPCX S-1 (via secondary aggregation) |
| 22 | 2026-Q1 | xAI quarterly revenue / operating loss | $818M / $2.47B | SPCX S-1 (via secondary aggregation) |
| 23 | 2026-07-07 | Congressional scrutiny of SpaceX DoD satellite-contract concentration (Space Data Network / Golden Dome) | flagged in reporting; no legislative action yet | SpaceNews (Jul 7), WebSearch corroboration |
| 24 | 2025-11-02 | Starcloud-1 launch — first Nvidia H100 GPU operating in orbit | 60kg satellite; trained NanoGPT (Karpathy) on Shakespeare corpus in orbit; predates this dossier's own creation | WebSearch (DataCenterDynamics, CNBC, NVIDIA blog, Space.com) — corrects prior "one-actor phenomenon" framing, see Superseded |
| 25 | 2026-06 (~) | Starcloud Series A | $170M at $1.1B valuation, led by Benchmark/EQT Ventures, oversubscribed | WebSearch aggregation |
| 26 | 2026-06-24 | "Orbital" (LA startup) FCC filing | 100,000-satellite / 10GW orbital-compute constellation proposal | WebSearch (SpaceNews/DCD/MLQ, reported Jul 1) |
| 27 | 2026-10 (planned) | Starcloud next launch | adds multiple H100s + Blackwell-platform GPU, reported plan | WebSearch (NVIDIA blog) |
| 28 | 2026-06-15/16 | DOJ intervenes in NAACP/Earthjustice/SELC Clean Air Act suit vs. xAI Colossus 2 gas turbines (Southaven, MS) | seeks dismissal on national-security grounds; no ruling yet | WebSearch (CNBC, Utility Dive) |
| 29 | 2026-06-30 | Starlink Memphis/Southaven residential price cut | 50% off, timed to xAI Colossus gas-turbine litigation PR fallout | WebSearch (Bloomberg) |
| 30 | 2026-07-01 (~) | SpaceX $25B bond secondary-market debut | trades weak out of the gate; Invesco (Brill) calls performance "very sloppy" | WebSearch (Bloomberg, Jul 2) |
| 31 | 2026-07-07 | Starlink Aviation price increase | Regional 25GB $2,000→$4,000/mo; Global Unlimited $10,000→$20,000/mo; hardware $145K→$200K; new "Aviation Regional Unlimited" tier $12,500/mo; effective Aug 7 | WebSearch (TechSpot, Yahoo Finance) |
| 32 | 2026-07-11/12 | SpaceX 30-year bond spread | tops 2.0pp over Treasuries (yield 6.7%→7.3%), wider than ~1.55-1.99pp average BB/junk spread despite Baa1/BBB rating | WebSearch (Seoul Economic Daily, BigGo Finance) — independently re-verified this pass |
| 33 | 2026-06-16 to 07-14 | SPCX equity | peak ~$225 (Jun 16) → ~$136.12 close (Jul 14), ~40% decline, within ~1% of $135 IPO price | WebSearch (Motley Fool, 247wallst) — closing price independently re-verified this pass |
| 34 | 2026-07-14 | MoffettNathanson initiates SPCX coverage | Neutral, $131 price target (below trading price); cites valuation "disconnect from quantifiable financial realities" | WebSearch (SatNews) |
| 35 | 2026-07-13 | BofA analyst (Ron Epstein) on AI1 | calls orbital data centers "one of the more speculative pieces" of the SpaceX investment thesis | WebSearch (Yahoo Finance/Reuters syndication) |
| 36 | 2026-07-14 | DoD Golden Dome new awards (not SpaceX) | $1.75B to L3Harris ($955M) and Sierra Space ($798M) | WebSearch (Washington Times) |
| 37 | 2026-07-16 | Starship Flight 13 pad abort | 4 of 33 Super Heavy Raptor 3 engines failed to ignite; 20 production Starlink V3 sats + first Starship-bay deployment scrubbed | WebSearch/SpaceNews (Foust), corroborated SpaceflightNow/NASASpaceFlight, Musk statement — independently re-verified this pass |
| 38 | 2026-07-19/20 | Starship Flight 13 second slip | re-targeted from Jul 20 22:45 UTC to Jul 23 22:45 UTC; 2 more Raptor 3 engines swapped, SpaceX says propulsion system "modified" | WebSearch (kfgo.com, teslarati, space.com — 5+ outlet cluster), independently re-verified this pass; outcome still pending as of writing |
| 39 | 2026-07-14 to 07-20 | SPCX equity daily close | $136.08 (14th) → $135.27 (15th) → $131.11 (16th) → $123.99 (17th, holding through 20th) — first sustained close below $135 IPO price | WebSearch (marketbeat.com daily-close table), independently re-verified this pass |
| 40 | 2026-06-26 to mid-07 | SpaceX 2056 bond tranche | spread 175bp (issuance) → 231bp; price ~99.45¢ (issuance) → ~90¢ on the dollar; yield ~7.5%; all 5 tranches now below issue price | WebSearch (Motley Fool, BigGo Finance, ZeroHedge — cross-corroborated) |
| 41 | 2026-06-18 | SpaceX credit ratings confirmed (all 3 agencies, investment-grade) | S&P BBB stable; Moody's Baa1 stable; Fitch BBB+ stable; S&P reportedly projects negative FCF through 2029 | WebSearch aggregation |
| 42 | 2026-07-16 | Piper Sandler initiates SPCX coverage | Neutral, $156 price target | WebSearch (Benzinga/marketbeat aggregation) |
| 43 | 2026-07 (~) | SPCX Street analyst consensus | 23 analysts, avg target ~$240 (range $62-$800) vs. ~$124 trading price — wide dispersion, no convergence | WebSearch (stockanalysis.com aggregation) |
| 44 | 2026-07-14 | Sen. Jack Reed (D-RI), SASC ranking member, on SpaceX | "longstanding concern about SpaceX's near monopoly position in launch and proliferated satellite constellations," SASC nominations hearing | WebSearch (armed-services.senate.gov hearing record, Payload Space) |
| 45 | 2026-06-12 to 2025-09 | Congressional SPCX stock purchases | Reps. Cisneros (D-CA), McGuire (R-VA), Meuser (R-PA), James (R-MI) ~$130K combined post-IPO; Rep. McClain (R-MI) up to $250K pre-IPO Sept 2025 private placement, shortly before a Pentagon-xAI deal; 4 of 5 sit on SpaceX-overseeing committees | WebSearch (NOTUS, CNBC Jul 3, Detroit Metro Times — cross-corroborated) |
| 46 | 2026-07-14 (~) | Reuters: xAI Colossus 2 unpermitted-turbine count | 59 confirmed (57 in Southaven, MS) — more than double the 27 xAI acknowledged Jan 2026; modeled NOx/CO/formaldehyde emissions near majority-Black communities | WebSearch (Reuters-sourced cluster: Utility Dive, TechTimes, TheNextWeb, US News) |
| 47 | 2026-06-15 | DOJ intervention filing detail | argues Colossus 2 supports US military operations "including operations involving Iran" as national-security grounds for dismissal | WebSearch (Reuters-sourced cluster) |
| 48 | 2026-07 (~) | Sen. Sheldon Whitehouse opens Senate investigation into xAI unpermitted-turbine pattern | separate from the DOJ Clean Air Act intervention | WebSearch (Reuters-sourced cluster) |
| 49 | 2026-03-13 | Starcloud's own FCC filing (distinct from its operational demo sat) | up to 88,000-satellite orbital-compute constellation, accepted for filing | WebSearch (SpaceNews) — corrects 2026-07-15 dossier undercount, see Superseded |
| 50 | 2026-03-19 | Blue Origin "Project Sunrise" FCC filing | up to 51,600 satellites, sun-sync 500-1,800km, tied to TeraWave broadband constellation + New Glenn | WebSearch (SpaceNews, DataCenterDynamics) — corrects 2026-07-15 dossier undercount, see Superseded |
| 51 | 2026-01-30 | AI1 FCC filing detail — compute-node pilot | pilot testing of on-orbit compute nodes on standard Starlink V3 hardware, planned "later in 2026" | WebSearch (SatNews summary of SAT-LOA-20260108-00016) — precedes dedicated AI1 prototype launch (~early 2027) |
| 52 | 2025-FY | SpaceX group revenue split by segment | $18.7B total: Starlink/connectivity $11.4B, launch and space work $4.1B, AI segment $3.2B (on a $6.4B AI operating loss) | Fortune (Jul 19 2026), reporting SPCX S-1 figures — first segment-level split carried by this dossier |
| 53 | 2026-05-20 | Anthropic-SpaceX compute contract (terrestrial Colossus 1 + 2) | $1.25B/mo, ~$15B/yr run-rate, ~325,000 Nvidia GPUs, 90-day termination either side; reported term through May 2029 (~$45B lifetime) | SPCX S-1 disclosure via Axios (May 20 2026) and Fortune (Jul 19 2026); monthly rate well corroborated, the term/lifetime figure is secondary-aggregation only |
| 54 | 2026-07 (~) | SpaceX-Pentagon data-center-capacity talks | reported "potentially worth billions"; explicitly NOT final, no contract signed | Fortune (Jul 19 2026) |
| 55 | 2026-07 (~) | Colossus complex scale | ~2M sq ft across the Memphis area (Colossus 1 + 2), ~1GW compute power, stated plan for 1M GPUs | Fortune (Jul 19 2026) |
| 56 | 2026-07-24 | Starship Flight 13 outcome | SUBORBITAL flight; 20 functioning Starlink V3 deployed from the payload bay (first non-simulator deployment), ~20 min of on-orbit checkout (RF + laser links confirmed), then demised on reentry — NOT an operational orbital insertion; booster landing burn under-ignited into a hard splashdown; ship survived water impact intact (a first) | SpaceNews (Foust, Jul 25), quoting SpaceX's Michael Nicolls and Dan Huot |
| 57 | 2026-07-24 | Starship next-flight plan | Flight 14 may attempt the first orbital Starship mission; Flight 15 to launch from Florida (Shotwell, stated June 2026) | SpaceNews (Jul 25) |
| 58 | 2026-07-15 | Starlink Global Roam plan withdrawn | $400/mo unlimited international land plan closed to new customers Jul 15 2026, existing service ends Aug 17 2026; separately Roam Unlimited international allowance cut from 60 to 30 days per trip | Mobile Internet Resource Center (rvmobileinternet.com), reporting Starlink's own plan-page changes |
| 59 | 2026-07-17 to 07-27 | SPCX equity daily close | $123.99 (17th) → $119.85 (20th) → $123.54 (21st) → $115.26 (22nd) → $118.24 (23rd) → $115.07 (24th) → $113.50 (27th); all-time low $110.85 intraday Jul 23 | stockanalysis.com daily-close table, live-verified this pass — corrects the prior row's "holding through 20th" (Jul 20 actually closed $119.85) |
| 60 | 2026-07-27 | SPCX vs IPO price and peak | $113.50 = ~16% below the $135 IPO price and ~50% below the ~$225 Jun 16 peak; ~-25% month-to-date | stockanalysis.com; Motley Fool (Jul 27) |
| 61 | 2026-08-04 | SPCX first-ever quarterly earnings — date confirmed | Aug 4 2026 after close, 4:30pm ET audio webcast, Q2 2026; first post-IPO report and expected first 10-Q | SpaceX IR (ir.spacex.com) via CNBC (Jul 21) and Motley Fool (Jul 27) — corrects this dossier's prior "~Aug 6, unconfirmed" |
| 62 | 2026-08-06 | SPCX first lockup tranche | ~911.5M shares (20% of restricted holdings) unlock on the 2nd trading day after earnings — ~$116B at announcement, ~$106B at Jul 27 prices; reported as the largest single lockup expiry on record | MLQ, Investing.com, Morningstar, Motley Fool (Jul 19-27) — cross-corroborated |
| 63 | 2026-08-31 to 10-25 | SPCX subsequent lockup tranches | 7% tranches on Aug 31, Sep 10, Sep 25, Oct 10 and Oct 25; Musk's controlling stake and select executive holdings remain locked to Jun 12 2027 | purepowerpicks/MLQ/Morningstar aggregation — tranche dates are secondary-aggregation, not read off the lockup agreement itself |
| 64 | 2026-07-24 | HSBC initiates SPCX coverage | Hold, $115 price target — the first initiation whose target sits essentially AT the trading price rather than far above or below it | gurufocus/marketbeat aggregation |
| 65 | 2026-07-07 | Goldman Sachs / UBS / Evercore ISI initiate SPCX | Buy $205 (Goldman), Buy $210 (UBS), Outperform (Evercore ISI) — the bull leg of the dispersion | gurufocus aggregation |
| 66 | 2026-07 (late) | SPCX Street consensus, updated | 34 analysts, average target ~$236.71 vs ~$113.50 trading price — coverage grew from 23 analysts without converging | S&P Global poll via stockanalysis.com |
| 67 | 2026-07-07 | SpaceX 2056 bond price | ~94.52¢ on the dollar at the Jul 7 close, against a 99.45¢ issue price | Motley Fool (Jul 10) — refines the "~90¢" figure this dossier carried on 07-20 |
| 68 | 2026-07 (early) | SpaceX average credit spread, whole curve | 1.62pp over Treasuries vs a 1.55pp BB (junk) average and a 0.92pp BBB average — the entire curve prices speculative-grade, not only the 30-year tranche | 24/7 Wall St (Jul 15), Seoul Economic Daily (Jul 12) |
| 69 | 2026-07-27 | SpaceX SEC filing status since IPO | live EDGAR: newest filing is an 8-K dated 2026-06-26; zero 10-Qs, zero new registration or offering filings — falsifier (b) untripped for a 2nd consecutive pass | SEC EDGAR CIK 0001181412, live fetch this pass |
| 70 | 2025-05-14 | China's first dedicated orbital-compute constellation launched | 12 satellites, "Three-Body Computing Constellation" (ADA Space + Zhejiang Lab), 744 TOPS each, 8B-parameter on-board model; part of a "Star-Compute Program" targeting 2,800 satellites / 1,000 POPS | SpaceNews (May 2025), CGTN, english.www.gov.cn — predates Starcloud-1 by ~6 months, see Superseded |
| 71 | 2026-07-18 | Shanghai "Xingshu Plan" first satellites launched | Shanghai Xingshu Tiansuan Space Technology; verification phase 2 compute + 12 edge-compute sats; commercial phase 50 + 100; final target ~1,000; pitched as hourly on-orbit compute rental to energy/shipping/finance users | china.org.cn (Jul 19), SCMP, eWeek — unveiled at the 2026 World AI Conference, Shanghai |
| 72 | 2026-05-14 | Cowboy Space FCC filing — "Stampede Data Center System" | up to 20,000 orbital-data-center satellites, dawn-dusk sun-sync shells 700-1,000km, launches from 2028; company had just raised $275M for rockets whose upper stages double as compute platforms | SpaceNews (May 2026), Via Satellite (May 15) — a 5th US filer, missed by this dossier's 07-20 four-actor count |
| 73 | 2026-01-11 | Axiom Space orbital data-center nodes | first two ODC nodes launched to LEO — operational hardware rather than an FCC constellation filing | Axiom Space releases, DataCenterDynamics |
| 74 | 2026-07-22 | FCC Part 100 "Space Modernization" order — treatment of AI1 | large 2026 orbital-data-center applications, including SpaceX's Jan 30 up-to-1M-satellite filing, are EXCLUDED from the new streamlined processing timelines and continue under legacy procedures | FCC Space Modernization Order (adopted Jul 22 2026); TechTimes (Jul 22), SpaceNews |
| 75 | 2026-07-22 | NSSL Phase 3 Lane 1 contract ceiling raise | $5.6B → $17B (+$11.4B, reconciliation + baseline funds); 2025-2034 mission estimate 60 → ~170; SpaceX Lane 1 task orders to date ~$1.65B across three awards | Breaking Defense (Hitchens, Jul 22), on-record SSC spokesperson |
| 76 | 2026-07-23 | Farrar 2030 Starlink forecast vs sell-side | Farrar $48B revenue / 46M consumer subs; Deutsche Bank $72B/82M; MoffettNathanson $73B/70M; RBC $101B/100M+; Morgan Stanley $121B/116M | tmfassociates.com, "Forecasting Starlink…" (Jul 23 2026) |
| 77 | 2026-07-23 | Farrar on terminal cost and D2D wholesale ARPU | terminals averaged over $300 to produce in 2025; Starlink Mobile wholesale ARPUs are "tens of cents rather than multiple dollars per customer"; that "won't provide a return on $20B of spectrum purchased from EchoStar" | tmfassociates.com (Jul 23 2026) |
| 78 | 2026-02-08 | Farrar on the Starship/Starlink-V3 schedule and the SSO gap | "Starlink V3 at scale on Starship" has stayed ~12 months out for nearly three years; identifies the sun-synchronous/Vandenberg-pad gap as the orbital-data-centre blocker | tmfassociates.com, "The problem with Starship" (Feb 8 2026) — back-catalogue item, first engaged by this dossier this pass |
| 79 | 2025-07 (~) | MDEQ position on the Colossus 2 turbines | Mississippi DEQ wrote to xAI that generators "affixed to a portable unit (i.e., a flatbed trailer)" can be treated as Clean Air Act "mobile" sources — the state-law basis xAI relies on against the permit claim | MDEQ letter via Mississippi Today; Earthjustice notice of intent (2026-02-13) |
| 80 | 2026-08 (late) | Colossus 2 preliminary-injunction hearing scheduled | a federal judge in the Northern District of Mississippi is set to weigh the NAACP/Earthjustice/SELC injunction request in late August 2026 | Energy.Media, Utility Dive — dates the "reported late-August, unconfirmed" note this dossier carried |
| 81 | 2026-07-20 | Falcon 9 Vandenberg pad abort | post-ignition automatic shutdown at T-0, SLC-4E, booster B1082, 24 Starlink V2 Mini (Starlink 17-39); no vehicle/pad/payload damage, next-day reflight; second SpaceX ignition anomaly in five days | Spaceflight Now (Jul 20), corroborated Space.com/Tech Times; brief 2026-07-21T0443Z |
| 82 | 2026-07-28 to 07-31 | SPCX equity daily close | $116.49 (28th) -> $112.55 (29th) -> $112.20 (30th) -> $108.37 (31st, new closing low) | investing.com + marketbeat.com daily-close tables, cross-corroborated live this pass |
| 83 | 2026-07-28 | SPCX new all-time intraday low | $107.01, on a day that ranged to a $118.13 intraday high and closed $116.49 — a volatility spike, not a smooth decline | investing.com historical data, live-verified this pass |
| 84 | 2026-07-31 | SPCX vs IPO price and peak, updated | $108.37 = ~20% below the $135 IPO price and ~52% below the ~$225 Jun 16 peak | derived from investing.com/marketbeat close, this pass |
| 85 | 2026-07-30 | Morgan Stanley (Adam Jonas) on SPCX | reiterates Overweight/$232; argues a sub-$100 print would mean the market is "undervaluing, or not valuing at all," SpaceX's AI efforts (Grok, Cursor), calls current levels "an attractive entry point" | The Motley Fool (Jul 29), quoting Jonas's note |
| 86 | 2026-07-29 to 07-30 | SpaceX NSSL Phase 3 Lane 1 draw-down | $1.6bn across two task orders for 18 Falcon 9 missions (Space Based Sensing and Targeting: Tracking Layer, Space Data Network Backbone, ABMTI), through end-2027 from Vandenberg; Space Force calls the ~2-month requirement-to-award cycle "unprecedented"; pushes SpaceX's 2026 Pentagon contract total past $8bn | SpaceNews (Jul 29, pubDate 2026-07-29T20:43:51Z), corroborated Breaking Defense same day; brief 2026-07-30T0343Z |
| 87 | 2026-07-31 (dated) | Starship Flight 14 target date | NET ~2026-09-30, Booster 21/Ship 41; may attempt first orbital insertion and first tower-catch of an orbital-class ship | spacelaunchschedule.com / nextspaceflight.com, WebSearch this pass — falls outside this pass's Aug 4/Aug 6 falsifier-testing window |
| 88 | 2026-07-27 to 08-01 | SpaceX SEC filing status, reconfirmed | live EDGAR: still no filing since the 2026-06-26 8-K as of Aug 1 — falsifier (b) untripped for a 3rd consecutive pass | briefs 2026-07-28 through 2026-08-01 (EDGAR watchlist, each cycle) |
| 89 | 2026-08-04 | SpaceX Q2 2026 total revenue (first post-IPO quarterly report, period ended Jun 30 2026) | $7.81B, +92% YoY (from $4.06B); beat ~$6.8-6.9B consensus by ~$1B | SpaceX Q2 2026 results via CNBC, Reuters/investing.com, CNN, Fortune — cross-corroborated; direct EDGAR/IR fetch 403-blocked |
| 90 | 2026-08-04 | SpaceX Q2 2026 group net loss / EPS | net loss $541M (narrowed ~$467M from a $1.0B Q2 2025 loss; vs a reported $4.3B Q1 2026 loss); EPS -$0.09 vs -$0.26 consensus (FactSet est net loss ~$1.9B) | @AlphaSenseInc quoting release text; CNN; CNBC — cross-corroborated |
| 91 | 2026-08-04 | SpaceX Q2 2026 adjusted EBITDA | $3.5B, +191% YoY (from $1.2B); vs ~$2.1B estimate | @AlphaSenseInc quoting release; CNBC/Fortune |
| 92 | 2026-08-04 | Q2 2026 segment — Connectivity (Starlink) | revenue $4.29B (vs $3.83B exp); operating income $1.66B — the only profitable segment | CNBC, Reuters/investing.com, CNN — cross-corroborated |
| 93 | 2026-08-04 | Q2 2026 segment — AI | revenue $2.56B (vs $2.18B exp); operating loss $1.26B (vs ~$2.39B est loss) — Colossus GPU-rental revenue offsetting xAI's losses | CNBC, CNN — cross-corroborated |
| 94 | 2026-08-04 | Q2 2026 segment — Space (launch) | revenue $962M (vs $835M exp); operating loss $542M — Starship dev load | CNBC, CNN — cross-corroborated |
| 95 | 2026-08-04 | SpaceX Q2 2026 capex | ~$18.37B total, AI ~$15.8B of it (from $7.7B AI capex in Q1) — ~$73B annualized; largest quarterly outlay disclosed. NOTE: single-cluster sourcing; consensus estimate disputed across outlets ($13.2B-$18.6B), so treat the beat/miss framing as unresolved | secondary reporting cluster (marketbeat/247wallst-adjacent) — softest-sourced line this pass |
| 96 | 2026-08-04 | Starlink subscribers, end-Q2 2026 | 12M across 164 countries (~2x YoY; +17% QoQ from ~10.3M end-Q1); Q2 net adds ~1.7M — reaccelerating vs Q1's 1.4M | Reuters/investing.com, CNBC (12M); Q1 10.3M base per S-1/Motley Fool preview |
| 97 | 2026-08-04 | Starlink ARPU, Q2 2026 | $66/mo, flat QoQ, down from ~$85 YoY — multi-quarter decline halted, NOT guided upward (direct falsifier (a) input) | Reuters/investing.com — cross-corroborated |
| 98 | 2026-08-04 | SpaceX Q2 2026 US government awards disclosed | "over $6 billion" in multi-year US government contracts cited in the earnings release | @AlphaSenseInc quoting release text |
| 99 | 2026-08-04 | SPCX equity — earnings-day reaction | regular session +9.43% to $125.33 (best day since Jun 15's +20%; anticipatory, ahead of the after-close release), then -5.27% after-hours to $118.72 on the actual numbers; still below the $135 IPO price | investing.com/Motley Fool (regular close); after-hours via CNBC/marketbeat-cluster |
| 100 | 2026-08-04 | SpaceX Q2 2026 10-Q — primary filing accession captured in the mirror | Form 10-Q accession 0001628280-26-052535 (plus the earnings 8-K 0001628280-26-052515), both filed 2026-08-04, captured by the 2026-08-04T20:50Z post-close source-mirror run — the index only (accession + primary-doc name); the filing body is not yet fetched (SEC live-fetch 403s this session) | sources/latest/edgar-spacex.json (CIK 0001181412), post-close snapshot |
| 101 | 2026-08-04 | SpaceX Q2 2026 10-Q — filing BODY now mirrored and read directly | sources/latest/edgar-docs/000162828026052535.htm (2.27MB, full inline-XBRL 10-Q) and 000162828026052515.htm (8-K cover) both fetched at HTTP 200 — the ops gap flagged last pass ("extend the mirror to fetch the filing document, not just the index") is closed | sources/latest/edgar-docs-index.json, this pass |
| 102 | 2026-08-04 | Q2 2026 revenue disaggregation by type — Connectivity | Consumer $2,485M / Enterprise & Government $1,806M (latter includes Starlink Mobile, not broken out separately) | PRIMARYSEC 10-Q, Note 3 (Revenue) — primary filing body read directly |
| 103 | 2026-08-04 | Q2 2026 revenue disaggregation by type — AI | Advertising (X) $367M / AI Solutions & Infrastructure $2,194M | SEC 10-Q, Note 3 |
| 104 | 2026-08-04 | Q2 2026 revenue disaggregation by type — Space | Launch Services $648M / Launch & Development $314M | SEC 10-Q, Note 3 |
| 105 | 2026-08-04 | Q2 2026 capex by segment, primary-sourced | Space $1,174M / Connectivity $1,367M / AI $15,828M / total $18,369M | SEC 10-Q, Note 18 (Segments) — confirms and resolves the prior pass's secondary-sourced/disputed capex line |
| 106 | 2026-08-04 | Q2 2026 D&A and share-based comp by segment | D&A: Space $158M / Connectivity $805M / AI $1,885M ($2,848M total); SBC: Space $179M / Connectivity $136M / AI $516M ($831M total) | SEC 10-Q, Note 18 |
| 107 | 2026-08-04 | Q2 2026 Segment Adjusted EBITDA, primary | Space -$205M / Connectivity $2,597M / AI +$1,146M (first positive quarter) / group $3,538M | SEC 10-Q, Non-GAAP reconciliation |
| 108 | 2026-08-04 | Q2 2026 group net loss and Adjusted EBITDA, primary-confirmed | net loss $541M; Adjusted EBITDA $3,538M (matches the prior pass's secondary-sourced figures almost exactly) | SEC 10-Q, Non-GAAP reconciliation |
| 109 | 2026-08-04 | SpaceX cash/liquidity position, Jun 30 2026 | cash & equivalents $93,522M + short-term marketable securities $6,487M (~$100B combined, matches CFO's on-call figure); $38,433M aggregate debt principal, no material maturity before Jul 15 2031 | SEC 10-Q, Liquidity and Capital Resources |
| 110 | 2026-08-04 | SpaceX H1 2026 cash flow | operating +$3,466M (positive, vs +$351M H1 2025); investing -$34,487M (capex $28,476M + $13,630M securities purchases, partly offset by $7,248M maturities); financing +$100,291M (IPO $85,675M + Notes/other proceeds) | SEC 10-Q, Consolidated Statements of Cash Flows |
| 111 | 2026-08-04 | SpaceX Cursor (Anysphere) acquisition | call option exercised June 2026, Agreement and Plan of Merger signed; $60B implied equity value, all-stock (VWAP-priced), expected close Q3 2026; concurrent GPU-compute collaboration agreement (Grok co-development) | SEC 10-Q, Note 20 (Acquisitions) |
| 112 | 2026-08-04 | SpaceX Mesh Optical Technologies acquisition | closed Jul 6 2026; ~3.8M Class A shares consideration, optical transceiver/data-center-interconnect maker | SEC 10-Q, Note 20 |
| 113 | 2026-08-04 | SpaceX customer concentration, Q2 2026 | "Customer A" (spans all 3 segments) 18.3% of consolidated revenue (~$1,430M; ~$809M implied in Q1); "Customer B" (AI segment only) 19.5% (~$1,524M, ~$0 implied in Q1) | SEC 10-Q, Note 3 — identities undisclosed; Customer A's all-segment pattern is consistent with, not confirmed as, the US government; Customer B's Q2-only ramp is consistent with, not confirmed as, the Anthropic Colossus contract |
| 114 | 2026-08-04 | SpaceX Valor equipment-lease financing (related party) | $2,039M current + $11,290M non-current debt on balance sheet (AI infrastructure hardware, treated as a "failed sale-leaseback"); $327M/$513M interest expense (Q2/H1); counterparty Valor Equity Partners, whose CEO Antonio Gracias is a SpaceX director | SEC 10-Q, Note 17 (Related Party Transactions) |
| 115 | 2026-08-04 | SpaceX litigation accrual, wiretapping-style class action | $354M accrued for probable/estimable losses (Electronic Communications Privacy Act, California Invasion of Privacy Act, CA constitution claims) | SEC 10-Q, Note 16 (Commitments and Contingencies) |
| 116 | 2026-08-04 | SpaceX-EchoStar spectrum deal, total consideration (primary) | ~$19.6B: ~$11.1B equity (261.8M shares at a fixed $42.40/share) + up to $8.5B EchoStar-debt payoff; $856M already paid this quarter under the spectrum credit agreement | PRIMARYSEC 10-Q, Liquidity and Capital Resources — refines Farrar's "$20B" figure (07-23) to a primary breakdown |
| 117 | 2026-08-04 | Starlink subscriber definition, backlog, deferred revenue (primary) | 12.0M subscribers excludes managed enterprise/government/aviation/maritime accounts; backlog $47,461M (56% expected <1yr, 34% 1-3yr, 10% thereafter); deferred revenue $14,286M (up from $12,116M at 2025FY-end); service availability in 167 markets | SEC 10-Q, MD&A and Note 3 |
| 118 | 2026-08-04 | SpaceX PP&E by category, Jun 30 2026 vs Dec 31 2025 | Servers/networking $34,771M (from $22,694M); Satellites $13,788M (from $11,949M); Machinery/equipment $9,453M (from $6,343M); Data center infrastructure $3,991M (from $2,960M); Launch sites $3,118M (from $2,404M) | SEC 10-Q, Note 5 |
| 119 | 2026-08-04 | AI segment turned Adjusted EBITDA-positive, CFO confirmation | "we turned adjusted EBITDA positive for the segment in the second quarter at $1.1 billion"; cloud-services agreements at Colossus/Colossus II contributed $1.6B of incremental AI revenue in Q2 | [INBOX] Space Exploration Technologies Corp._Earnings Call_2026-08-04_English.md (Bret Johnsen, CFO) |
| 120 | 2026-08-04 | SpaceX CFO guidance — AI compute payback and ARR | "less than a 1-year payback" on new AI compute capex; additional $6.7B of cloud-services revenue contracted in the first weeks of Q3 (ramping from Oct, 6-month term); guided $100B+ ARR by Dec 2026 (incl. Cursor contribution) | [INBOX] earnings call transcript (Bret Johnsen, CFO) |
| 121 | 2026-08-04 | SpaceX nameplate compute draw | 1.4GW at Jun 30 2026 (from 1.0GW Q1 2026, 0.4GW Jun 2025); guided "over 2 gigawatts" by end-2026 | SEC 10-Q (Key Business Metrics) and earnings call, cross-confirmed |
| 122 | 2026-08-04 | Musk internal $1T revenue target pulled forward | "reaching $1 trillion in revenue" moved from 2031 to 2030 ("non-zero chance" of 2029) — revenue, not ARR | [INBOX] earnings call transcript (Elon Musk) |
| 123 | 2026-08-04 | Starlink Mobile terrestrial "femtocell" plan, verbatim detail | Shotwell: base stations mounted on Starlink dish hardware ("little femtocells... around the country"), explicitly avoiding upfront low-band spectrum spend; Musk: "small stations... just Starlink dishes that also provide connectivity in the mobile spectrum bands" | [INBOX] earnings call transcript — sharpens priors.md's 08-05 femtocell note to verbatim quotes |
| 124 | 2026-08-04 | Orbital-compute satellite renamed/specified on-record | Musk: "Starmind AI satellite," an "optimized Vera Rubin NVL72 computer," targeting launch "next year" [2027] | [INBOX] earnings call transcript (Elon Musk) — sharpens this dossier's AI1 tracking |
| 125 | 2026-08-05 | SPCX equity daily close | $108.27, -13.61% (new closing low; "second-worst day on record" per press) | investing.com / stockanalysis.com, cross-corroborated live this pass |
| 126 | 2026-08-06 | SPCX first lockup tranche, outcome | ~911.5M shares became eligible (tradable float 4.9%→11.8% of shares outstanding, ~639M→~1.55B shares); stock closed $114.92, +6.14%, on 252.4M shares of volume (109% above the 3-month average of 121.0M); no Form 144/4 cluster confirming large-scale insider selling found — falsifier (b)'s cleanest test yet, UNTRIPPED | Motley Fool, CNBC, tradingkey.com, stockanalysis.com — cross-corroborated this pass |
| 127 | 2026-08-07 | SPCX equity daily close | $133.11-$133.27 (two independent close tables within $0.16), +15.83-15.97%, on the Terafab announcement and an Argus Buy/$160 upgrade; ~1.5% below the $135 IPO price | investing.com / stockanalysis.com, cross-corroborated live this pass |
| 128 | 2026-08-06/07 | SpaceX/Tesla "Terafab" chip complex, announced | $16.8B first phase (potentially up to $119B total across phases), Grimes County, TX; single facility for logic/memory/packaging/test; chips for Optimus, robotaxis, and orbital data centers; 100M+ sq ft planned, 3,000+ jobs targeted | Electrek, TechCrunch, Yahoo Finance/qz.com (Aug 6-7 2026), cross-corroborated — Tesla's specific co-financing role not yet independently confirmed beyond the joint announcement |
| 129 | 2026-08-07 | Argus initiates/upgrades SPCX | Buy, $160 price target (analyst Steven Silver), citing rapid AI-compute payback | Benzinga (Aug 7), WebSearch aggregation |
| 130 | 2026-03 (adopted) | China 15th Five-Year Plan (2026-2030) — commercial-space/ODC policy context | drafted Oct 2025 by NDRC, formally adopted March 2026 at "Two Sessions"; Jan 2026 CASC + Chinese Society of Astronautics conference produced 3 (unpublished) planning documents referencing "gigawatt-level space digital infrastructure"; a possible 3rd Chinese ODC actor, Nayuta Space ("ALAYA" constellation concept), also surfaced | ESPI Brief 77 "Constellating Compute" ([INBOX] ESPI-Brief-77-Constellating-Compute.md) — reinforces, does not change, the two-bloc Superseded correction |
| 131 | 2026-08-04 | SpaceX Q2 2026 capex guidance, forward | CFO Johnsen: "the next 2 quarters are very similar to the current quarter from a CapEx level perspective" — ~$18B+/quarter confirmed continuing into Q3/Q4 2026, not moderating | [INBOX] earnings call transcript (Bret Johnsen, CFO) — mined further this pass |
| 132 | 2026-08-04 | SpaceX compute-capacity target, sharpened | Musk: "tentative target" of 20GW power/cooling online by end of 2027, "probably" ~15GW realistic (allows ~25% of projects to slip); ~10% of total compute reserved for internal Grok training, remainder for lease; monetization guess "somewhere between $30 and $50" per watt | [INBOX] earnings call transcript (Elon Musk) — mined further this pass |
| 133 | 2026-08-04 | SpaceX H1 2026 operational detail | 78 total launches / 1,041 tons to orbit H1 2026; constellation ~10,200 satellites at Jun 30 (9,600 broadband, ~800 Tbps downlink); new Q2 Starlink Mobile international carrier partners: SoftBank, NTT Docomo, Spark New Zealand | [INBOX] earnings call transcript (Gwynne Shotwell) — mined further this pass |
| 134 | 2026-05-12 | FCC approves SpaceX-EchoStar spectrum assignment (DA 26-471, GN Docket 25-302) | ~65MHz (AWS-4, AWS-H Block, unpaired AWS-3) assigned via a 2-step transaction; confirms Shotwell's Aug 4 on-call "recently approved" claim; EchoStar conditioned to fund a $2.4B escrow/trust within 30 days of consummation; step-two final consummation (SpaceX holding the licenses outright, vs. the current interim Spectrum Business Trust 2025-1) targeted Nov 30, 2027 | PRIMARYFCC DA 26-471 (primary text read in full this pass, docs.fcc.gov) — resolves Open Question 17; full deal mechanics tracked in dossiers/mss-2ghz-spectrum.md, not restated here |
| 135 | 2026-08-11 | SpaceX beneficial-ownership disclosure — Antonio Gracias / Valor Equity Partners | Schedule 13G: 503,414,530 Class A shares (6.5%), shared voting/dispositive power only (no sole power), event date 2026-06-30, filed as a passive investment; same Gracias/Valor entities hold the $13.3B "failed sale-leaseback" AI-infrastructure equipment-lease debt (10-Q Note 17 above); Gracias is a SpaceX director | SEC EDGAR Schedule 13G, accession 0001213900-26-087780, filed 2026-08-11 |
| 136 | 2026-08-11 | Starlink subscriber/Mobile-user disclosure, internal all-hands video (posted publicly) | 13M+ consumer subscribers (from 12M at the Jun 30 10-Q close); 22M monthly Starlink Mobile users; 12M Direct-to-Cell connections; ARPU $66/mo confirmed flat; ~11,000 satellites in orbit, 167 countries | PRIMARY@SpaceX X video, 2026-08-11T21:04:14Z; corroborated basenor.com, TeslaNorth (Aug 12), @JackKuhr (TIER-B mirrored handle) — primary video itself unfetchable from this sandbox, treat as company-disclosed-not-independently-verified per brief 2026-08-12T0350Z |
| 137 | 2026-08-11 | Musk guidance: AI segment revenue to exceed all other SpaceX segments combined | "probably next month" [September 2026], "will significantly exceed all other SpaceX revenue in Q4" — forward guidance from an unscripted internal setting, not a filing or analyst call; not yet confirmed in a Q3 disclosure | SpaceX internal all-hands video via TeslaNorth (Aug 12) — single/secondary-sourced, not independently corroborated beyond the same video |
| 138 | 2026-08-10 | SPCX equity — reclaimed IPO price | closed above the $135 IPO price for the first time since Jul 15, 2026 (+3.76-4%) | CNBC (Aug 10), tradingkey.com (Aug 10) — cross-corroborated on the move/framing; exact closing price not pinned this pass given cross-source disagreement (see Analytic Stance) |
| 139 | 2026-08-10 | tmfassociates.com, Farrar: "Musk's daily Starship problem" | disputes Musk's "≥1 Starship flight/day within a year" and 100K+-satellite ambition as demand-unjustified: at ~10K subs/V3-satellite, the 15-20K satellites Shotwell has separately guided to already implies 150-200M global subscribers, near the total addressable fixed-broadband market; identifies terminal manufacturing (Bastrop capacity to ~300K/week, ~12-13M/yr by year-end) rather than launch cadence as the binding near-term constraint — implies a ceiling of ~2M net adds/quarter, in line with, not below, the pace already observed; also flags fiber competition ($50/mo symmetric 10Gbps vs. Starlink's gateway-cost economics near $75K/Gbps/month) | tmfassociates.com, "Musk's daily Starship problem…" (Aug 10, 2026) |
| 140 | 2026-08-12 | SpaceX Cursor (Anysphere) merger — closing status, as of this pass | not yet closed; no 8-K filed on EDGAR since the Aug 4 10-Q (checked through Aug 12); one no-byline SatNews report claims regulatory review "approaching completion," close expected "before end of August 2026" and cites "$10B in deferred service commitments" — UNCONFIRMED, that figure not found in the primary 10-Q Note 20 text, treated with the desk's standing SatNews caution | EDGAR live check this pass (CIK 0001181412); SatNews (Aug 12, no byline), not independently corroborated |
| 141 | 2026-07-30/31 (~) | xAI/SpaceX Colossus 2 temporary-turbine removal agreed order, Mississippi DEQ | 69 unpermitted turbines to be fully removed by Jul 2027 (extended from a Dec 31 2026 MDEQ deadline), removal beginning Aug 2026; replaced by a permanent 1.2GW/41-turbine gas plant operating under a Clean Air Act permit granted Mar 2026 — a genuine desk gap, never logged by this dossier or any brief through this pass despite predating it by ~3 weeks | TechCrunch (Jul 31), Engadget, Gizmodo, qz.com, Tom's Hardware — cross-corroborated, WebSearch this pass |
| 142 | 2026-07-29 (~) | Rep. Frank Pallone (D-NJ), House Energy & Commerce ranking member, letter to Musk/SpaceXAI | demands full document production (turbine inventory, emissions/permitting records, official communications incl. with the White House) and a Colossus site visit for committee staff, deadline Aug 11, 2026; no confirmed SpaceXAI compliance or response found via live search this pass | House E&C Democrats letter (democrats-energycommerce.house.gov, PDF); CNBC (Jul 29), qz.com — also a prior-pass miss |
| 143 | 2026-08-24 (scheduled) | NAACP/Earthjustice/SELC v. X.AI Corp. preliminary-injunction hearing, N.D. Miss. (3:26-cv-00074) | evidentiary hearing on the Clean Air Act shutdown motion confirmed for Aug 24, 2026 — sharpens this dossier's prior "late August 2026, unconfirmed" placeholder to a specific date; no ruling as of this pass (5 days out) | CourtListener docket 3:26-cv-00074, cross-corroborated named reporting, WebSearch this pass |
| 144 | 2026-08-14 | SpaceX Cursor (Anysphere) acquisition — CLOSED, primary mechanics | Merger effective Aug 14, 2026; 389,289,254 Class A shares issued via a 7-day VWAP pricing formula (implies ~$154.13/share = $60.0B ÷ 389,289,254 shares, desk calc — the first time SpaceX's own paper has disclosed an implied per-share valuation); 1,752,426 additional shares for vested Cursor RSUs; ~29.1M RSUs and ~44.4M options assumed and converted; issued under the Section 4(a)(2) private-placement exemption — resolves Open Question 10 | SEC 8-K, accession 0001628280-26-056945, filed 2026-08-14T08:23:02Z |
| 145 | 2026-08-13/14 | SpaceX new Schedule 13G beneficial owners | Founders Fund/Peter Thiel entities 5.5% (427.3M Class A shares); Google LLC/XXVI Holdings/Alphabet 7.2% (551.2M shares), both event-dated 2026-06-30, passive, shared voting/dispositive power only; separately Musk's own 13G/A re-file (48.4%, routine, no stake change) | SEC EDGAR Schedule 13Gs, accessions 0001426012-26-000004 and 0001193125-26-352289 (both Aug 14), 0001104659-26-095936 (Musk, Aug 13) |
| 146 | 2026-08-14 | MoffettNathanson on SpaceX's terrestrial "femtocell"/small-cell wireless plan | assessed as "far from competitive" as a standalone network — coverage/propagation limits, spectrum scarcity (SpaceX holds roughly a fifth of what AT&T/T-Mobile/Verizon each hold), permitting/zoning/equipment-replacement friction; the real "bull case" is using the buildout as leverage to force one of the Big Three into an MVNO deal, all three of which have so far refused — first named-analyst skeptical read on the plan this dossier flagged 08-05 as forward-looking intent | Light Reading (Baumgartner, Aug 14), corroborated Fierce Network same day |
| 147 | 2026-08-10 to 08-19 | SPCX equity daily close, post-lockup week | $138.74(10) $133.29(11) $146.15/$146.27(12, two sources) $141.29(13) $140.00(14) [wknd] $146.23(17) $143.34(18) $139.65(19) — closed above the $135 IPO price on 6 of 8 trading days, the first full post-lockup week held mostly above IPO price | stockanalysis.com and marketbeat.com daily-close tables, cross-corroborated live this pass |
| 148 | 2026-08-14 | Starlink Direct-to-Cell — first African commercial launch | Airtel Africa + Starlink launched Africa's first commercial D2D mobile service in the DRC (Kinshasa announcement), compatible LTE Android phones, no special hardware, capped at SMS/light-data (WhatsApp); 30-day free trial; 14-market continent-wide rollout named, no committed timeline; follows a Mar 2026 Kenya pilot | PRIMARYSpace in Africa (Aug 16), Spacewatch Global (Aug 18), 9+ outlet corroboration; no primary Airtel/SpaceX release located |
| 149 | 2026-08-12/13 | Poland forces SpaceX to reverse a Starlink European-roaming exclusion within ~24h | SpaceX had excluded Polish customers from Starlink's EU roaming zone (~Aug 12), reversed after FM Sikorski threatened to reconsider the ~$50M/yr Poland pays for Ukraine's Starlink service; PM Tusk confirmed the reversal Aug 13; SpaceX characterized the original exclusion as an error | europeanspaceflight.com (Aug 18 write-up), brief 2026-08-19T0343Z |
| 150 | 2026-08 (undated) | SpaceX's own EU 2 GHz MSS lobbying position paper — Starlink Mobile European business-model detail | Starlink Mobile will NOT offer a standalone consumer service in the EU — wholesale-only via MNO partnerships (cites an existing Deutsche Telekom agreement across "ten countries"), targeting 2028 commercial D2D launch; SpaceX asserts its 2025 EchoStar spectrum acquisition carries global ITU priority rights that transfer "regardless of the EU's decision" — regulatory/spectrum mechanics owned by [mss-2ghz-spectrum](dossiers/mss-2ghz-spectrum), not restated here | [INBOX] SpaceX MSS EU White Paper_August 2026.md, exact publication date unverified |
| 151 | 2026-04 (~) | Farrar/TMF, via The Information: Starlink's segment profitability may partly be a transfer-price artifact | "If Starlink was paying full price for its launches, then it would be making a loss, even though it's an amazing business" — argues Connectivity's disclosed operating income rests on internal Falcon 9 launch costs below external market rate, not a market-tested transfer price; consistent with this project's own viability-fixture treatment of SpaceX's internal launch $/kg as un-anchorable (data/viability/starlink.json) — a pre-IPO critique never previously engaged by this dossier, surfaced this pass in the disconfirming-evidence hunt | The Information (Apr 2026), via @theinformation on X, WebSearch this pass |
| 152 | 2026-08 (desk calc) | Implied H2 2026 capex vs. Street revenue consensus | ~$36.8B capex (2 quarters at the ~$18.4B/quarter Q2 pace CFO guidance says holds) vs. ~$22.9B Street-consensus H2 revenue (~$9.7B Q3 + ~$13.2B Q4, implied by ~$39.9B FY2026 consensus) — capex would run ~1.6x H2 revenue under the company's own capex guidance and the Street's own top-line assumptions, before any Cursor-related capex layers on | desk calculation from SEC 10-Q Q2 capex ($18,369M, [tracked series](#tracked-series)) and Street FY2026 consensus (WebSearch aggregation, stockanalysis.com/futuresearch.ai — not company-guided) |
| 153 | 2026-08-14 | Q2 2026 13F wave — NVIDIA's SpaceX stake, primary | 122,764,805 Class A shares, $20,975,594,582 fair value at Jun 30 2026 (SpaceX's 2nd-largest disclosed holding in NVIDIA's own 13F, ~33% of that portfolio); position originated as NVIDIA's ~$10B piece of xAI's Jan 2026 $20B Series E, converted into SpaceX Class A stock when SpaceX acquired xAI (Feb 2026, [tracked series](#tracked-series)) — a chip *supplier* to Colossus is simultaneously a top-3 SpaceX equityholder | PRIMARYSEC Form 13F-HR, NVIDIA Corp, CIK 0001045810, accession 0001045810-26-000065, filed 2026-08-14 — primary informationTable.xml read directly this pass |
| 154 | 2026-08-14 | Q2 2026 13F wave — Saudi PIF's SpaceX stake, primary | 154,146,835 Class A shares, $26,337,528,228 fair value at Jun 30 2026 — 69.5% of PIF's entire disclosed US equity portfolio, PIF's first-ever disclosed SpaceX position; corrects secondary reporting's conflicting $21.5B (thesauditimes.net, priced off a stale $139.50/share) vs. $26.38B (satelliteprome.com) figures to the filing's own fair-value mark (~$170.9/share, consistent with SPCX's ~mid-$160s-$170s June 2026 trading range) | PRIMARYSEC Form 13F-HR, Public Investment Fund, CIK 0001767640, accession 0001011438-26-000537, filed 2026-08-14 — primary informationTable.xml read directly this pass |
| 155 | 2026-08-14 (~) | Q2 2026 13F wave — additional disclosed SpaceX stakes, secondary | AMD ~3.3M Class A shares (~$445-565M depending on valuation date, cost basis/acquisition date undisclosed); Harvard Management ~12.94M shares / $2.2B (>50% of Harvard's own disclosed US equity portfolio) — both are 13F-only disclosures this pass did not independently re-fetch from EDGAR (lower priority than the NVIDIA/PIF primary reads above) | CNBC, Bloomberg, Benzinga, Motley Fool, TipRanks (WebSearch aggregation, cross-corroborated) |
| 156 | 2026-08-20 | SpaceX second post-IPO lockup tranche | ~319M Class A shares became eligible for trading on SpaceX's 70th day as a public company; stock fell to ~$132.69-$134 (two sources disagree by ~$1.30), -4.05% to -4.9%, back below the $135 IPO price — a materially weaker absorption than the Aug 6 first tranche's same-day +6.14% ([tracked series](#tracked-series)); next tranches Sep 9, Sep 24, Oct 9 and Oct 24, full 180-day lockup expires Dec 8 2026, Musk's stake locked to Jun 2027 — corrects the prior pass's purepowerpicks/MLQ-sourced "Aug 31/Sep 10/Sep 25/Oct 10/Oct 25" tranche-date estimate | Benzinga, Yahoo Finance, Seeking Alpha, Schwab Network (WebSearch aggregation, cross-corroborated) |
| 157 | 2026-08-26 | SPCX equity daily close | $141.10, +1.05% — the most recent close independently pinned this pass; Aug 21-25 closes ranged an estimated $137-140 per aggregator commentary but were not pinned to individual days at the same confidence bar this dossier otherwise holds | WebSearch aggregation (live pass), not cross-checked against a second daily-close table at the row level |
| 158 | 2026-06-15 to 2026-08 (~) | Starlink $10/mo equipment-rental fee — introduced, then reversed | $10/mo hardware-rental fee added to all new US Residential-tier orders Jun 15, 2026; removed from new purchases by ~mid-July 2026 (about a month later); the free Mini second-terminal perk for Residential Max (pulled in the same June wave) was restored by early-to-mid August 2026 | rvmobileinternet.com, satelliteinternet.com, notebookcheck.net, BGR (WebSearch aggregation, cross-corroborated) — resolves a Farrar (Aug 23, X) question; corrects the trajectory this dossier's own 2026-07-27 CHANGELOG line called a "3rd consecutive monetization tightening" on this specific data point |
| 159 | 2026-08-21 (filed) | Colossus 2 Clean Air Act preliminary-injunction hearing, N.D. Miss. (3:26-cv-00074) — POSTPONED | the Aug 24, 2026 evidentiary hearing this dossier tracked as confirmed (previous pass) did not occur; a court notice filed Aug 21, 2026 postponed it, with no replacement date found this pass | AOL/Yahoo (SpaceXAI hearing-postponement reporting), WebSearch this pass |
| 160 | 2026-08-18 to 08-21 | Bernstein (analyst Doug Harned) raises SpaceX price target to $248, reiterates Overweight | argues orbital-data-center compute pricing is "absolutely central" to SpaceX's long-run AI thesis, models >$600B annual group revenue by 2031; separately calls Musk's "1 Starship launch/day by end-2027" claim achievable "if reuse works as planned" — above Harned's own base-case estimate, i.e. a bull case still gated on execution, not a restatement of the claim as fact | Advanced Television (Aug 21), TipRanks, 24/7 Wall St (Aug 18-21), cross-corroborated |
| 161 | 2026-08-25 | SpaceX + Louisiana Gov. Landry announce "Starbase Louisiana" megasite | $100B, Vermilion Parish; construction starts 2027, first launch NET 2029, 12+ towers targeting 30 Starship flights/day at full build-out, initial hiring 3,000+ (promotional estimate floats to ~10,000) — the largest capital investment in Louisiana history; same day SpaceX confirmed the Aug 25 Cape Canaveral Falcon 9 Starlink launch was the last Florida Starlink mission flown on Falcon 9, ending ~260 F9 Starlink missions from the Cape since May 2019 — all future Florida Starlink launches move to Starship | PRIMARYCNBC, Space.com, USNews, Fox8, KSLA, mynews13, Investing.com (6+ outlet corroboration; no SpaceX/Governor's-office primary release recovered) |
| 162 | 2026-08-21/22 | NVIDIA invests $250M in Starcloud (orbital-compute operator), commits dedicated "Space-1 Vera Rubin Module" hardware | Series A extension, total Starcloud raised ~$450M at a $2.3B valuation, co-led by Manhattan West with Cisco Investments also participating; NVIDIA's module is claimed ~25x H100-equivalent in-space compute — a backer/hardware-supplier commitment to an existing US orbital-compute operator, not a new bloc entrant; doesn't change the two-bloc field's composition but is the most material capital datapoint into orbital compute since the Aug 6-7 Terafab announcement | SpaceNews, Via Satellite, GeekWire, BusinessWire (Aug 21), cross-corroborated |
| 163 | 2026-08-14 to 08-27 | SpaceX SEC filing status, reconfirmed | live EDGAR direct fetch (submissions JSON, this pass): newest filing remains the Aug 14, 2026 cluster (Cursor-closing 8-K + two Schedule 13Gs) — zero new 8-K/10-Q/S-1/13G-13D in the 13 days through Aug 27 — falsifier (b) untripped for a further pass on an unchanged EDGAR watermark | SEC EDGAR CIK 0001181412, live fetch this pass |
Earlier claims that evidence disproved — kept on record. 5 corrections documented.
Believed 2026-07-05: this dossier's STATE OF PLAY described AI1 (the orbital compute satellite program) as having "reported anchor contracts of ~$1.25B/month (Anthropic) and ~$920M/month (Google) — call it ~$2.17B/month if both are real committed offtake," filed under the "orbital compute" subsection and treated as AI1's own revenue base. Overturned 2026-07-09: both contracts are confirmed (DataCenterDynamics, TechCrunch, CNBC, named reporting, Jun 2026) to be for capacity at SpaceX/xAI's *terrestrial* Colossus 1 GPU cluster (Memphis, TN) — an existing, already-operational data center, not the orbital AI1 satellites. AI1 itself remains pre-revenue: two prototypes targeting launch ~early 2027, zero signed commercial contracts (Anthropic has only "expressed interest" in future orbital capacity, per reporting). The dollar figures were correct; the attribution to orbital compute was not. This was a real analytical error, not a source error — the underlying reporting (e.g. TechCrunch's Jun 5 Google piece) was clear that the deal covered "xAI data centers," and the prior dossier version should have caught the terrestrial/orbital distinction at creation.
Believed 2026-07-05 through 2026-07-09: this dossier's STRUCTURAL DYNAMICS stated orbital compute "remains a one-actor phenomenon at the *orbital* layer specifically (no rival has an orbital data-center program at AI1's stage)." Overturned 2026-07-15: Nvidia-backed Starcloud has operated an H100 GPU in orbit (Starcloud-1, a 60kg satellite) since 2025-11-02 — predating this dossier's own 2026-07-05 creation — training small models (NanoGPT) in orbit, and raised a $170M Series A (Benchmark/EQT Ventures) at a $1.1B valuation; a second startup, "Orbital," filed with the FCC for a 100,000-satellite/10GW orbital-compute constellation on 2026-06-24. Neither is at AI1's claimed scale (up to 1M sat data centers) and neither has AI1's rumored hyperscaler-anchor ambitions, but the "one-actor" claim was factually wrong from the dossier's own creation date forward — a research gap in the original pass (an "is this true" search was never run on the claim itself), not a genuinely new development this pass. Per priors.md's promotion rule this doesn't cross the >=2-independent-blocs threshold (Starcloud and Orbital are both US actors, same bloc as SpaceX), so no first-class thread promotion is triggered — but the dossier's internal claim required correction.
Believed 2026-07-15: this dossier's own correction (above) characterized orbital compute's non-SpaceX field as two funded rivals — Starcloud (operational H100 demo satellite + $170M Series A) and "Orbital" Compute Inc. (100,000-satellite FCC filing, Jun 24). Overturned 2026-07-20: the field was already at least twice that size when the 07-15 correction was written. Starcloud separately filed its own, larger FCC application for up to 88,000 satellites, accepted for filing 2026-03-13 — a scale-up distinct from its small operational demo satellite and missed by 07-15 despite predating it by four months. Blue Origin filed "Project Sunrise" with the FCC on 2026-03-19 — up to 51,600 satellites, tied to its TeraWave broadband constellation and New Glenn launch capacity — a fourth, well-capitalized US actor also missed 07-15 despite the same four-month head start and public reporting (SpaceNews, DataCenterDynamics) well before that pass. Per priors.md's promotion rule this still doesn't cross the >=2-independent-blocs threshold (all four actors — SpaceX, Starcloud, "Orbital," Blue Origin — are US-bloc), so no first-class thread promotion is triggered, but this is the second consecutive pass in which this dossier undercounted the orbital-compute field on its own core topic — a research-method gap (finding one rival and stopping), not a new event either time. See Analytic Stance for the generalized lesson.
Believed 2026-07-15 through 2026-07-20, asserted twice in the two entries above: that orbital compute's actors were all US-bloc, and therefore that priors.md's ">=2 independent competing actors across >=2 blocs" promotion threshold was NOT crossed. Also asserted, from 2026-07-15: that Starcloud-1 (2025-11-02) was the first non-SpaceX orbital computer. Both overturned 2026-07-27. China's "Three-Body Computing Constellation" — 12 dedicated orbital-compute satellites built by ADA Space with Zhejiang Lab, each rated 744 TOPS and carrying an 8-billion-parameter on-board model — launched on a Long March 2D from Jiuquan on 2025-05-14, six months before Starcloud-1 and fourteen months before this correction, as the first phase of a "Star-Compute Program" targeting 2,800 satellites and 1,000 POPS. Shanghai Xingshu Tiansuan Space Technology launched the first satellites of its separate "Xingshu Plan" on 2026-07-18 (verification phase 2 compute + 12 edge-compute satellites; roadmap to ~1,000), explicitly selling hourly on-orbit compute. Both were covered in English at the time (SpaceNews, CGTN, gov.cn, SCMP). The US count was also short by two: Cowboy Space filed for a 20,000-satellite "Stampede Data Center System" on 2026-05-14 after a $275M raise, and Axiom Space has had orbital data-center nodes in LEO since 2026-01-11. The consequence is not cosmetic: the >=2-blocs threshold is crossed, so the promotion question priors.md's rule exists to answer is now live, and the field's operational leader by time-in-orbit is Chinese, not American. Third consecutive pass correcting this dossier's count on its own core topic — and the first in which the error was a bloc-level blind spot rather than an arithmetic one. See Analytic Stance.
Believed 2026-07-28 through 2026-08-04, asserted in STATE OF PLAY, POLITICAL & REGULATORY and ANALYTIC STANCE across five passes: that this project's source mirror had "no SpaceX/SPCX EDGAR entry" (CIK 0001181412), and — this pass — that "adding SPCX to the EDGAR mirror is now the highest-value ops fix." Overturned 2026-08-04 on direct inspection: an ops-repair added the CIK to scripts/source-mirror.sh on 2026-07-28, and sources/latest/manifest.json records the edgar-spacex fetch at HTTP 200 (the deep-dive-time snapshot, 2026-08-04T00:01Z, predated the after-close 10-Q; the post-close 20:50Z run then captured it — 10-Q accession 0001628280-26-052535). The mirror was never the gap. The actual constraints, narrower: the desk's live-fetch path 403s on SEC's declared-User-Agent policy (and a sandbox curl 403/000s on top of that), and the mirror stores the submissions *index* rather than the filing document. The "not mirrored" claim had been carried, uninspected, since the CIK was in fact added — a verify-before-citing miss on the dossier's own tooling. Corrected in ANALYTIC STANCE; scorecard OPS handoff rewritten accordingly.