Elon Musk was in his prime ahead of SpaceX's first earnings call as a public company, warning detractors of complete annihilation before the call, while throwing out trillion-dollar buzzwords with alacrity during the call itself.
SpaceX has now disclosed its earnings for the second quarter of 2026, reporting revenue of $7.8 billion against a consensus estimate of $6.93 billion, with Starlink contributing $4.3 billion and SpaceXAI $2.6 billion to this headline number. Of course, despite all of the hype surrounding SpaceX, it's still a loss-making entity, reporting a net loss of $541 million against a consensus estimate of -$1.94 billion.
It was SpaceX's earnings call, however, that offered the real fireworks. First, as we noted in a separate post, Elon Musk's star enterprise will exclusively use NVIDIA GPUs from here onward, with the Vera Rubin platform now expected to power its upcoming data centers.
Elsewhere, Musk disclosed that SpaceX's AI1 satellites will leverage NVIDIA's Space-1 Vera Rubin module , which consists of four Rubin GPUs, two Vera CPUs, and a hefty amount of memory. Do note that the AI1 satellites can support up to 150kW of peak compute payload, replete with liquid radiators, meteoride shielding, a centralized compute module, and deployable solar arrays. These satellites will be manufactured at SpaceX's Gigasat facility in Texas.
Coming back, SpaceX will bring around 2 gigawatts of compute online by the end of 2026, while targeting "closer to 10 than 5" gigawatts of compute by the end of 2027, replete with a tentative 20 gigawatt of power-and-cooling load.
Of course, SpaceXAI has already inked a number of cloud deals. According to SpaceX's CFO, these deals have a sub-one-year payback. What's more, the company has already contracted $6.7 billion of cloud deals in the first weeks of Q3, and intends to continue renting out its compute capacity, especially as just 10 percent of SpaceXAI's compute goes to train newer iterations of the Grok AI model, with the rest either reserved for inference or slated to be rented out.
Elon Musk reserved some of his most interesting comments for the ongoing memory crisis, noting:
"Limiting factor currently is memory. Memory output is increasing by around 20% per year. Now normally that would be fantastically fast and amazing for any large mature industry. But ask yourself is the demand increasing by 20% a year? No, the demand is increasing by 200% a year, maybe higher. So if you've got demand increasing much faster than supply then Economics 101 would suggest that the price increases. It does not decrease."
Elsewhere, Musk noted that SpaceX's revenue - not ARR - will jump to $1 trillion by 2030, a full year earlier than what was previously projected, while also "expecting to reach $100 billion plus ARR in December of this year."
Musk also turned his guns towards the Big Three US carriers - AT&T, Verizon, and T-Mobile - declaring that Starlink Mobile was aimed directly at them. Of course, SpaceX currently lacks terrestrial spectrum in sufficient quantities to start its own mobile service in urban areas. And, with these carriers refusing to enter into an MVNO with SpaceX, the company now wants to establish small, low-power femtocell-type cellular base stations on existing Starlink dishes to avoid the massive upfront CapEx required for building cell towers and acquiring the appropriate spectrum. We've already explained how SpaceX intends to make FCC spectrum auctions a very costly endeavor for these carriers.
Finally, Musk says that SpaceX is now gunning for a flight cadence of one Starship launch per day by 2027, with boots on the moon slated for 2028.
Follow Wccftech on Google to get more of our news coverage in your feeds.