$100T, one gigawatt at a time

The target as a ratio

$100T over 13,581,063,709 shares is $SPCX = ~$7,363, against $144.75 today. That is about 50.9x. Spread over 10 years it is 1.481x a year, so about 48% compounded (our estimate).

Three numbers turn that into a business: revenue x margin x multiple.

All rows are our estimate. The 45% is the long term GAAP net income margin target in SpaceX's IPO roadshow, which the company says is a goal, not a projection. Q3 revenue was $14.575B, or about $58.3B a year. On the call Elon put annual recurring revenue at $500B to $1T by the end of 2027 (unofficial transcript; the Motley Fool summary matches). From there, $7.41T needs about 25% to 35% growth a year for nine years (our estimate).

Unit one: a gigawatt

On the call, the CFO said pricing is "tracking slightly above the high end of the $30 to $50 per watt range" (unofficial transcript, checked against Motley Fool). SpaceX has not said whether that is a yearly price; the table treats it as yearly. Nameplate compute was 1.7 GW at 30 September, up from 0.6 GW a year earlier, per the Q3 release.

All values are our estimate. Elon said "I would be surprised if we brought on less than 13 gigawatts of compute by the end of next year" (unofficial transcript, checked against Motley Fool). From 13 GW to 370 GW is about 28x in nine years, about 45% a year. The jump from 1.7 GW to 13 GW is 7.6x in 15 months (our estimate).

Unit two: a Starship launch

The CFO, in the interview SpaceX filed with the SEC: "It's roughly 200 launches for every gigawatt that we put up. About 5 megawatts per launch." That is "the first generation of the satellite and the rocket," he added.

Steps 2 to 5 are our estimate. Step 4 uses the prospectus goal to cut launch cost "by 99% or more" from the $18,500 per kg historical average, times 100 metric tons, our assumption for a full Starship V3. It leaves out satellite and chip cost, which SpaceX does not disclose.

Unit three: a subscriber

Starlink had 13.6M consumer subscribers at $69 a month in Q3, about $828 a year each. $7.41T at $828 is about 9 billion subscribers, about 660 times today (our estimate). Great business, wrong unit. The watt gets there. The dish does not.

The base rate

Nvidia was worth $57.5B at the end of 2016 and about $5.12T now, per CompaniesMarketCap. That is about 89x in under 10 years, about 58% a year. Apple did about 8.3x over the same stretch, about 24% a year (our estimate). So 48% a year has been done, by the company selling the picks for the AI build. SpaceX starts about 34 times larger than Nvidia did. The pace has a precedent. The starting size does not.

The second base rate is SpaceX's own. Per the prospectus, citing NASA, Falcon 9 cut launch cost to about $2,700 per kg in 2010, 85% below the $18,500 average, and Falcon Heavy reached about $1,400 in 2018. Starship's goal is another 93% drop from Falcon 9 (our estimate).

The equivalence

$100T is about 78% of the IMF's $128.0T of world output for 2026, and about 62% of the $161.7T it projects for 2031. It is about 19.5 times Nvidia, the largest company today, and about 4 times the seven largest companies combined, SpaceX included (our estimate from CompaniesMarketCap on 23 October).

The physical check

The prospectus states the plan in plain numbers: "We aim to launch 100 gigawatts of AI compute capacity on solar-powered satellites each year, equivalent to roughly one fifth of total annual U.S. power production in 2025." It says that takes "thousands of launches per year" and "approximately one million metric tons to orbit annually." At 100 t a flight that is 10,000 flights a year, about 27 a day (our estimate). At the first generation 5 MW a launch it is 20,000.

Elon on the call: "We expect Starship to fly weekly at some point in the first half next year" (unofficial transcript, checked against Motley Fool). Weekly to 27 a day is about 190x. Hard, but it is a factory problem, and the prospectus says the fully reusable Starship positions SpaceX to launch that level of mass.

The second order effect

On Earth, power is the wall. Elon said "power cooling, electrical equipment infrastructure is somewhat of a limiting factor," and that SpaceX will aim for "an incremental 20 gigawatts of power cooling and infrastructure next year" (unofficial transcript, checked against Motley Fool). The US generated 4.43 thousand TWh in 2025, per the EIA, an average of about 506 GW (our estimate). For scale, our base case has about 296 GW of SpaceX compute running by 2036, near 60% of that, though most of it would be in orbit on solar power.

Once compute moves to orbit, the price of a watt is set by the price of a kilogram to orbit, so every cut in launch cost becomes a cut in compute cost. The rocket stops being a product and becomes the cost curve of the AI business.

Our base case

Every line below is our estimate. Value is 30 times each line's earnings.

Launch97

Compute in orbit69

Grok, X and Grok Bot25

Terafab chips16

Ground compute10

Starlink6

Net margin is about 48.7%. On this path $100T is first crossed in 2033 (our estimate). By 2036, $100T would be only about 13.4 times earnings. Our weighted view, 30% bull at $591T, 50% base at $224T and 20% bear at $19T, is about $293T. Speculation, not a price target.

What would make this wrong

The bear case is short. Orbital compute fails to ship at scale and launch stays a niche market. Remove both lines and the base falls to about $58T, still about 30 times today (our estimate).

No compute satellite has flown at the scale this needs. The $1B a flight launch price assumes customers pay about $5,000 per kg on a 200 t future ship, well above Falcon 9's $2,700, so the two biggest lines rest on markets that do not exist yet.

Not financial advice.