SpaceX’s August 4 earnings report is framed as the proof point: $7.81 billion revenue, up 92% y/y, Starlink subscribers doubled to 12 million, and the company still missed on a $6.93 billion estimate by nearly $1 billion while cutting an expected $2 billion net loss to about $0.5 billion. The bigger thesis is that SpaceX’s three pillars — connectivity, AI, and space — are already monetizing fast enough to fund an extreme capex program, including almost $18.5 billion of capex, a $50 billion signed backlog, and a push toward trillion-dollar revenue by 2030. Short term, the stock is treated as a live lockup/unlock trade with $100 billion of shares going live and 25% to 36% of supply unlocking in the next couple of months; long term, the argument is that if the execution holds, the stock can reprice violently higher. The strongest catalyst chain is the Nvidia partnership: exclusive Blackwell/Vera Rubin-era buildout, a claim of the largest GPU arsenal and data-center cluster on Earth, and a path to 10 to 20 gigawatts of AI infrastructure by 2030.
August 4 was treated as SpaceX’s prove-it day because the first earnings report had been expected to fail under a bearish Wall Street view that Elon Musk could not execute a money-vacuum plan.
SpaceX posted $7.81 billion revenue, up 92% y/y, versus a $6.93 billion estimate, and the company’s Starlink base doubled to 12 million subscribers.
Despite the blowout top line, the stock was still down the morning after earnings, with the move blamed on tomorrow’s unlock of $100 billion of SpaceX stock.
The report still showed a net loss of about $0.5 billion, but that was a sharp improvement from the expected $2 billion loss; EPS was expected at -$0.26 and came in at $0.09.
The company’s three pillars were broken out as connectivity, AI and space, with AI up 250% y/y and space up 30% y/y.
Elon Musk said SpaceX will have trillion-dollar revenue by 2030, moved up from 2031, and the claim was framed as 40% or 32% above the biggest company on Earth today.
The stock was described as facing major pressure because 911 million shares are unlocking, equal to about $100 billion of potential sell-side supply, with 25% to 36% of the float expected to unlock over the next couple of months.
Capex was said to be just under $18.5 billion, 39% above expectations and nearly matching NASA’s roughly $25 billion annual budget.
A $50 billion backlog of signed revenue was cited as already booked because SpaceX can build data centers quickly enough to collect contracts others cannot.
SpaceX’s AI stack was described as a major customer of Nvidia, with Elon Musk saying SpaceX will build exclusively on Nvidia because Vera Rubin is the best architecture and the best AI computer.
The company was said to be doubling its GPU fleet by the end of next year, with scenarios laid out for 2 to 2.5 gigawatts by year-end, 3 gigawatts by the end of this year, 6 gigawatts by end-2027, and 10 to roughly 20 gigawatts by around 2030.
Jensen Huang’s rule of thumb was cited as 40 to 50 billion dollars of annual revenue per gigawatt, implying 15 gigawatts could mean 600 to 750 billion dollars of revenue by the end of the decade before adding space and connectivity.
The segment claimed SpaceX is exclusive to Jensen and Nvidia, has the largest GPU arsenal and largest data-center cluster on Earth, and is selling AI compute to Anthropic and Google.
Starlink v3 was described as launching as soon as next year on the Starship launch sequence, with B3 satellites to be deployed, turned on, and made to work after orbit insertion.
Gwynne Shotwell was cited as targeting a Verizon/AT&T-like network for most of the roughly 5.5 billion internet users worldwide, plus direct-to-cell coverage that would remove dead zones and make iPhones broadly connected next year.
The company was described as one of the most shorted names in the market, with Tesla used as the analogy: the most shorted stock for 2 to 3 years before a massive rerating and a 13x move in 2020 to 2021.