SpaceX stock: how much of tomorrow is already in the price?

SpaceX’s Starlink, enterprise and government “Connectivity” segment produced $4.291 billion of Q2 revenue: $2.485 billion from Consumer and $1.806 billion from Enterprise & Government. Operating income was $1.656 billion, or 38.6% of revenue by Buteon calculation; capital expenditure was $1.367 billion.

That is the strongest current support for the valuation. Revenue grew 65.8% and operating income 79.4% from the prior-year quarter. A profitable network with expanding customer groups is different evidence from a launch program still in development.

But two popular metrics cover only part of the segment. The filing counts each Starlink service line as a subscriber and excludes managed enterprise and government contracts. Average monthly revenue per subscriber—ARPU—was $66, down a company-reported 22.4%. Neither metric covers all Connectivity revenue.

Segment operating income is not cash generation. It includes depreciation and does not supply a segment cash-flow statement. The durable question is whether growth, satellite replacement, ground infrastructure and terminal economics produce cash after the capital needed to keep the network competitive.

Primary source: SpaceX Q2 revenue, metrics and segment tables (opens in a new tab).

The launch business contains an accounting trap. SpaceX recognizes revenue from customer launches, but no intersegment revenue when Falcon launches its own Starlink satellites; those costs are capitalized into the satellites. Internal launch volume can strengthen the network without creating Space-segment revenue.

Cadence did not rise in the comparable filing period. Q2 Falcon launches fell to 37 from 45, as internal launches fell to 27 from 36; customer launches edged up to 10 from 9. H1 Falcon launches were 77 versus 81, and Starship flew once versus three times.

Flight 13 on 24 July released twenty Starlink V3 test satellites on a suborbital path. All twenty re-entered; this was not operational deployment. The ship completed a controlled water landing. Ten of thirteen booster landing engines restarted, only five appeared to remain lit at impact, and the booster was lost in a hard splashdown.

That flight retired real engineering questions. It did not establish orbital deployment, booster recovery, repeated reuse or cost per flight. Starship becomes economic evidence when technical milestones repeat often enough to change launch cost, customer revenue or the capital required to deploy Starlink—not when a single test produces an impressive clip.

Primary sources: SpaceX Flight 13 mission record (opens in a new tab); CBS Flight 13 outcome report (opens in a new tab); SpaceX Q2 launch metrics (opens in a new tab).

SpaceX’s AI segment extends beyond Grok to X advertising, subscriptions and API access, data licensing, cloud services and infrastructure. Recast comparisons include businesses combined under common control.

Q2 AI revenue was $2.561 billion: $367 million of advertising and $2.194 billion of AI Solutions & Infrastructure. The segment lost $1.257 billion from operations and recorded $15.828 billion of capital expenditure—86.2% of total reportable-segment capex by Buteon calculation.

The filing adds a useful measurable: AI Nameplate Compute Draw rose to 1.4 gigawatts from 0.4 gigawatts a year earlier. It means installed GPUs multiplied by their all-in power draw. SpaceX explicitly says it is not actual electricity use or utilization, so it measures capacity built, not capacity economically absorbed.

The next proof is not simply more compute. It is whether revenue and operating economics catch the installed-capacity curve without requiring financing costs to compound just as quickly. That is the same discipline applied more broadly in Buteon’s evidence review of the AI capital cycle.

Primary source: SpaceX Q2 AI segment and key metrics (opens in a new tab).

Elon Musk’s 13 August Schedule 13G reports 6,418,547,515 Class A-equivalent shares and 48.4% of its denominator, with a 30 June event date. It is an ownership report and records no August purchase.

About four-fifths—5,116,475,230 shares, or 79.7%—reconciles to the June Form 3 after preferred-stock conversions, including 350 million options exercisable within 60 days.

The remaining 1,302,072,285 restricted Class B shares—20.3%—were mentioned in the Form 3 remarks but expressly excluded from its reported holdings. The 13G is neither a new purchase nor simply the old reported total.

The SpaceX CEO Award covers 1,000,000,000 shares and depends on market-cap, service and permanent million-person Mars-colony conditions. The AI CEO Award covers 302,072,285, with distinct market-cap, service and 100-terawatt non-Earth data-center conditions.

The 48.4% calculation counts issued Class A plus Musk’s Class B equivalents, restricted shares and options. This person-specific denominator is neither whole-company ownership nor voting power; adding those securities to both sides lowers the percentage.

Primary sources: Musk Schedule 13G (opens in a new tab); Musk Form 3 and remarks (opens in a new tab); Raw Schedule 13G submission (opens in a new tab).