SpaceX Q2 Earnings Call Targets $1 Trillion Revenue and a Fourth U.S. Carrier

Key Takeaways

During the SpaceX Q2 earnings call, management said new cloud contracts and rapid AI infrastructure expansion could lift annual recurring revenue above $100 billion by the end of 2026. Elon Musk moved the company’s $1 trillion annual revenue target forward to 2030, with 2029 presented as an outside possibility. SpaceX also outlined plans to combine satellites with a terrestrial network and compete directly with AT&T, Verizon and T-Mobile in a market management valued at approximately $600 billion, with President Gwynne Shotwell predicting that Starlink Mobile could attract a significant number of their customers.

SpaceX held its first earnings call as a publicly traded company on August 4 after reporting second-quarter revenue of approximately $7.8 billion, up 92% year over year and more than 10% above Wall Street forecasts.

Adjusted EBITDA increased approximately 191% to $3.5 billion, beating the roughly $2 billion consensus by 75%. SpaceX reported a loss of $0.09 per share, considerably smaller than the projected loss of $0.23 to $0.26. (SpaceX Q2 earnings call transcript)

SpaceX Moves Its $1 Trillion Revenue Target Forward

Executives provided limited conventional financial guidance, but Chief Financial Officer Bret Johnsen said SpaceX could finish 2026 with annual recurring revenue of at least $100 billion. Cloud-computing contracts are expected to provide much of the increase.

The forecast is considerably higher than the company’s current quarterly revenue run rate and depends on contracted AI capacity becoming operational during the remainder of the year.

Musk also moved SpaceX’s longer-term revenue target forward by one year. The company is now aiming to generate $1 trillion in annual revenue by 2030 instead of 2031. Musk said reaching that level in 2029 was not impossible, although Wall Street estimates remain far below that projection.

The target relies heavily on SpaceX’s AI compute business. Musk estimated that AI infrastructure could produce between $30 and $50 of annual revenue for each watt of deployed computing capacity. Maintaining that revenue rate while expanding capacity into multiple gigawatts would be necessary for the company to approach its trillion-dollar ambition.

High spending will continue in the near term. Johnsen said capital expenditure during the third and fourth quarters would probably remain close to the second quarter’s $18.4 billion level.

Starlink Mobile Takes Aim at America’s Largest Carriers

The most significant new strategic announcement involved Starlink Mobile, a planned consumer wireless service that would compete more directly with AT&T, Verizon and T-Mobile.

Shotwell described the three carriers as businesses representing approximately $600 billion in combined annual revenue. She said SpaceX expects its service to attract many of their customers because it could provide broader coverage, fewer dead zones and stronger performance during natural disasters.

Starlink Mobile is expected to begin offering services by the end of 2027. SpaceX plans to start launching mobile-focused satellites using 65MHz of EchoStar spectrum and supplement the satellite network with terrestrial infrastructure.

Management acknowledged that a full mobile service requires more than satellites. However, Musk argued that SpaceX would not need to copy the traditional carrier model of constructing large, expensive cellular towers.

Instead, the company plans to deploy smaller base stations that could be added to Starlink terminals installed on homes and commercial buildings. Those rooftop terminals already have unobstructed views of the sky and broadband connections, potentially allowing them to provide local connectivity to mobile phones on the ground.

The proposal would turn parts of Starlink’s installed terminal base into a distributed terrestrial network. Whether the approach can match conventional carriers on capacity, reliability and regulatory requirements remains unproven, but it could reduce the capital needed to enter the mobile market.

Starlink added a record 1.7 million subscribers during the second quarter, taking its total customer base to 12 million. Average revenue per user remained approximately $66.

Management said next-generation Starlink V3 satellites could offer around 10 times the performance of V2 models. Starship is also expected to carry much larger satellite batches, creating the potential for a substantial increase in network capacity and revenue. Musk said it was possible that Starlink could eventually provide the majority of the world’s internet connectivity.

AI Expansion Locks SpaceX Into Nvidia’s Ecosystem

SpaceX spent approximately $15.8 billion on AI computing infrastructure during the second quarter, accounting for most of its $18.4 billion in total capital expenditure.

The company’s operational computing capacity increased from approximately 1GW at the end of the first quarter to 1.4GW. Management expects it to exceed 2GW by the end of 2026.

For the end of 2027, Musk said operating compute capacity could reach between 5GW and 10GW and would probably be closer to the upper end of that range. He separately floated a provisional goal of installing enough power, cooling and electrical equipment to support as much as 20GW. The larger figure appears to describe infrastructure readiness rather than the company’s formal forecast for active computing capacity.

Musk said SpaceX had decided to build its future AI systems exclusively around Nvidia technology because it considered Nvidia’s architecture the strongest available. He expects SpaceX to receive a substantial proportion of Nvidia’s GPU supply next year. The announcement also ruled out AMD hardware for the company’s planned infrastructure.

SpaceX plans to use Nvidia’s Vera Rubin NVL72 rack-scale systems in terrestrial data centers and co-design computing payloads with Nvidia for orbital deployment. The first Starmind AI satellites are expected to launch in 2027, beginning SpaceX’s effort to place data-center capacity in orbit.

Johnsen said current AI infrastructure investments have a payback period of less than one year. SpaceX’s AI compute operation generated positive adjusted EBITDA of approximately $1.1 billion during the second quarter, its first positive result on that measure.

Musk argued that SpaceX’s experience building reusable rockets gives it an engineering advantage in data-center construction. In his view, applying even part of the company’s rocket-manufacturing expertise to power, cooling and computing infrastructure should allow SpaceX to expand more efficiently than conventional operators.

Grok Roadmap Accelerates as Cursor Deal Nears Completion

The July launch of Grok 4.5 drove a threefold increase in token consumption, according to Shotwell. SpaceX expects that usage to continue rising as it releases new versions of the model.

Musk said Grok 4.6 was likely to launch the week after the earnings call, followed by Grok 4.7 approximately three to four weeks later. Grok 5 remains scheduled for release before the end of 2026.

The company plans to train Grok 5 using engineering information accumulated during SpaceX’s nearly 25-year history. Musk predicted that incorporating data from rocket development, manufacturing, satellite operations and launch systems would make Grok significantly more capable at engineering tasks.

Musk also confirmed that SpaceX’s acquisition of Anysphere, the company behind the Cursor AI coding platform, was close to completion. The all-stock transaction was announced at approximately $60 billion and is intended to strengthen SpaceXAI’s coding and enterprise software capabilities.

Executives did not address speculation about a possible combination between SpaceX and Tesla during the call.

Musk Says Starship’s Heat-Shield Problem Is Solved

Musk offered an optimistic assessment of Starship’s technical progress, saying SpaceX now believes it has resolved the spacecraft’s heat-shield problem. A reusable thermal-protection system has long been considered one of the largest obstacles to rapid Starship reuse.

SpaceX plans to attempt the first tower catch of a Starship upper stage during Flight 14, expected later in August. The mission is also set to deploy the first operational V3 Starlink satellites and may attempt the recovery of both the Super Heavy booster and the spacecraft.

Musk projected that Starship could reach a flight frequency of at least once per day in roughly one year. That remains an ambitious target because the vehicle is still in flight testing and has not yet demonstrated routine full reuse.

Shotwell said the Artemis III mission planned for 2027 would test docking between Orion and commercial lunar-lander vehicles in low-Earth orbit. NASA’s current schedule then targets the Artemis IV mission for a crewed lunar landing in 2028.

Musk said Falcon rockets currently deliver approximately 2,500 metric tons into orbit annually, representing an estimated 80% to 90% of global orbital payload mass. SpaceX’s long-term objective is for Starship to increase that figure to more than 1 million tons per year and eventually as much as 10 million tons.

The scale of those plans helped explain SpaceX’s elevated spending, but it also contributed to investor caution. Shares fell more than 8% in after-hours trading as the market weighed the company’s growth targets against the capital required to fund AI infrastructure, Starlink Mobile and Starship simultaneously.