SpaceX's Real Profit Engine Isn't the Rocket — It's the Bandwidth

A Raptor engine quit mid-ascent and the mission was cut short, but all 26 Starlink V3 satellites deployed on Flight 14 are working flawlessly.

A Raptor vacuum engine quit mid-ascent, the mission got cut from a planned ten hours to roughly three, and yet the headline number that matters came through clean: all 26 Starlink V3 satellites deployed during Monday's Flight 14 are working flawlessly. That combination — a hardware hiccup that didn't stop the payload — captures exactly where SpaceX's investment story now sits.

The rocket lifted off at 8:46 a.m. local time from Starbase in Texas. When the engine failed, flight controllers chose to press on rather than abort. About 26 minutes after liftoff, the vehicle completed its orbital-insertion burn using just one remaining engine, then after two laps around Earth began its braking maneuver roughly two hours and twelve minutes into the mission. The ship splashed down under control in the North Pacific near Hawaii. For the first time, the giant vehicle reached low Earth orbit.

A schedule problem, not a demand problem

The shortened flight carries consequences beyond the test card. Reuters reported Tuesday that the incident could add further strain to NASA's tightly drawn timeline for crewed lunar landings under Artemis. More than $15 billion has already been sunk into developing the reusable transport system — a reminder of how capital-hungry space hardware remains. The space division's second-quarter 2026 results make the point bluntly: $962 million in revenue against an operating loss of $542 million.

Full and rapid reuse of both stages is the precondition for pushing launch costs down to a quarter of today's level. Until that works, Starship stays a formidable cost center. But it is no longer the center of the investment case.

Bandwidth is the business now

Each V3 satellite delivers roughly ten times the downlink capacity of the previous generation, and the 26 units released Monday add about 26 terabits per second to the Starlink network, according to company figures. Starlink closed the second quarter of 2026 with 12 million subscribers — double the count a year earlier. Bernstein estimates consumer and small-business broadband revenue could reach about $64 billion by 2031.

The sharper shift runs toward artificial intelligence. Analyst John Blackledge projects that leasing terrestrial AI compute to customers such as Google and Anthropic — not the traditional launch business — will become the company's largest revenue contributor by the first quarter of 2027. Data-center capacity earmarked for that push is set to climb from 2.1 gigawatts in 2026 to 6 gigawatts by the end of 2027.

Government money as ballast

Long-term public contracts underpin the commercial build-out. On September 18, NASA modified an existing agreement with the company by $946 million, adding three more crewed flights to the International Space Station and extending astronaut missions through 2030. The total value of the arrangement rises to $5.92 billion across 17 flights.

Analysts see pricing power holding up alongside it. Mizuho Securities said Friday that SpaceX can defend its premium rates for compute capacity into 2027, while media reports on Thursday pointed to favorable ratings driven by AI growth momentum and an attractive risk-reward profile.

What the tape is saying

The equity traded at EUR 131.50 in European dealing Tuesday, up 2.7% on the day, and sat about 7.7% above its 50-day average of EUR 119.10 in premarket. Against the 52-week high of EUR 194.46, the shares remain 32% below their peak — a discount that reflects the heavy upfront costs of the transition.

Judging this company by the yardsticks of conventional aerospace no longer works. The rockets are the visible spectacle; the economics are being decided where orbital data infrastructure meets warehouse-scale computing.