The Second Space Age Is Bigger Than SpaceX

SpaceX reported its third quarter on October 20, and the headline numbers were the kind that get a stock talked about: revenue of $14.6 billion, up 161% from a year earlier, net income of $4.4 billion, and its first profitable quarter since the June IPO.¹ The second space age finally had a profit attached to it. Then the segment table came out, and the rocket company turned out to be something else.

Of that $14.6 billion, $7.6 billion came from the Super Intelligence segment, the AI business, and $5.2 billion came from Starlink connectivity. The Space segment, the rockets, brought in $1.8 billion, including $1.13 billion of launch services, and posted an operating loss of $7 million. Capital spending for the quarter was $19.6 billion, and $16.4 billion of it went to the AI segment. Nameplate computing power reached 1.7 gigawatts at the end of September, up from 1.4 gigawatts in June.¹ By where it spends its money, the rocket company is a data center company with a launch division.

That is the fact this report is built on. The money going into space is real and growing, but the biggest name in the industry has just said, in its own filings, that most of its revenue and almost all of its new investment is going somewhere else. If the second space age is a market rather than a company, the question changes: not whether to own the famous rocket, but who gets paid when the customers, the rivals and the suppliers all show up at once.

How We Got Here: From a Race to a Market

The first space age was a contest between two governments, and the money went to a handful of prime contractors picked in Washington. The second one began when the launch price fell and the customer list widened. The World Economic Forum and McKinsey, in a report published in April 2024, sized the global space economy at $630 billion in 2023 and projected it would reach $1.8 trillion by 2035, growing at almost twice the rate of global GDP. They also said the share captured by incumbent space hardware and service providers would gradually decrease, and that six industries, among them supply chain, state-sponsored defense and digital communications, would generate more than 60% of the increase.²

The first stretch of that forecast has run ahead of schedule. The Space Foundation’s midyear report put the global space economy at $686 billion in 2025, up 12% in a year, with a compound annual growth rate of 9.8%. Commercial activity was $544.3 billion of that, 79% of the total, while 53 government space budgets added up to nearly $141 billion, of which U.S. government funding of $78.3 billion was 57%. Orbital launch value rose 33.5% to $19.17 billion, and the report projects military space spending will top $150 billion in 2027.³

Three events in 2026 turned those numbers into a story the public market could trade. On April 10, Artemis II splashed down off San Diego after a nearly ten-day trip around the Moon and back, the first crewed lunar voyage in more than half a century, with Reid Wiseman, Victor Glover, Christina Koch and Canada’s Jeremy Hansen aboard.⁴ In June, SpaceX went public at $135 a share and closed its first day at $160.95.⁵ And on September 28, Starship Flight 14 reached orbit about 26 minutes after leaving Starbase in Texas and released 26 Starlink V3 satellites, even though it lost an engine shortly after liftoff and a planned ten-hour mission was cut to three hours and nine minutes.⁶ SpaceX called it the first revenue-generating Starship mission.¹

A crew around the Moon, a public rocket company, an orbital Starship: the second space age in three headlines. What they do not show is where the biggest company’s money is actually going.

What SpaceX’s Own Numbers Say

The third quarter was not a one-off. SpaceX’s second quarter, reported on August 4, showed revenue of $7.81 billion and a net loss of $541 million, with capital expenditure of $18.37 billion, of which $15.83 billion went to the AI segment. The space unit took in $962 million and lost $542 million at the operating line. The stock fell more than 8% in late trading on the spending figure, leaving it more than 40% below the peak it set four days after listing.⁷

By the third quarter the AI segment had grown 195% from the prior quarter to $7.6 billion, helped by the Cursor acquisition, which closed on August 14 at an implied equity value of $60 billion. Nine-month capital spending reached $48.1 billion. Starlink ended the quarter with 13.6 million consumer subscribers, up from 12 million in June, paying an average of $69 a month across 174 countries, served by about 11,100 satellites in orbit.¹

The launch business, meanwhile, is not the thing growing. SpaceX flew 35 missions in the quarter, 12 for customers and 23 for itself, down from 44 a year earlier, and 113 launches so far this year. The Space segment’s operating loss narrowed from $542 million to $7 million, with the losses attributed to Starship research and development.¹ Read as a whole, the filings describe a connectivity company that owns a launch division and is building data centers at a pace that dwarfs both.

✓ Trusted partner One trader who followed SpaceX from its early days says the biggest opportunity of the second space age is a company that is not SpaceX, and he names it →

The market has traded the stock accordingly. It closed at $108 on August 5, the day before the first insider unlock, and ended October 12 at about $163, roughly 50% higher, through six unlocks totaling about 2.3 billion shares. The largest, up to 1.3 billion shares, became eligible on the second full trading day after the third-quarter report, and another 797.6 million shares unlock after the close on December 8. Elon Musk’s own shares are locked until June 2027. About 21% of the 13.6 billion shares outstanding were tradable in mid-October, a figure expected to approach 40% by early December.⁵

None of this makes SpaceX a bad business. It makes it a different business from the one the phrase “space stock” describes: an investor buying it for rockets and landers is mostly buying AI data centers and a broadband subscription.

The Customers Are Spreading the Money

Start with NASA. On Friday, October 23, the agency held what it called an Artemis Acceleration briefing at Kennedy Space Center. Artemis III remains on track for summer 2027, but it will not land on the Moon. It is a 14-day crewed rehearsal in low Earth orbit built around three launches: a Blue Origin Blue Moon Mark 2 test lander on New Glenn, four astronauts on SLS and Orion, and a SpaceX Starship test vehicle. Orion docks with Blue Moon on day three and with the nose of Starship on day eight. NASA still plans to put a crew near the lunar south pole on Artemis IV in 2028, with a possible second landing on Artemis V the same year, and it has not decided whether Starship or Blue Moon will carry that crew.⁸

That structure is the point. In February NASA restructured the program, moving Artemis III up a year and turning it into a lander test flight whose results will decide which lander carries astronauts on Artemis IV. The agency is paying for two landers from two companies and keeping both in the race. Its timeline has an uncrewed Starship landing on the Moon before the end of 2027 and Blue Origin’s uncrewed Mark 2 landing attempt in the first quarter of 2028.⁹ The ship-to-ship propellant transfer demonstration is set for early 2027, and the Aerospace Safety Advisory Panel has estimated about 22 tanker launches to fill a Starship lander, against the six to eight Musk has argued for.⁸

The Space Force runs the same play. In October 2025 it assigned the first batch of missions under Phase 3 Lane 2 of the National Security Space Launch program, the lane for must-go payloads, to fly no earlier than fiscal 2027: five launches to SpaceX for $714 million and two to United Launch Alliance for $428 million. Blue Origin, the third Lane 2 provider, received none because New Glenn had not yet flown the second mission it needs for certification. The service anticipates 54 launches under Lane 2 in all.¹⁰ On May 28, 2026, the Space Force gave Blue Origin a Lane 1 task order for a National Reconnaissance Office mission from Cape Canaveral, targeted for late 2027 or early 2028; SpaceX, ULA and Blue Origin are the three providers eligible in that lane.¹¹

Then there is Golden Dome. In April the Space Force named 12 companies to build prototypes of space-based interceptors for the missile shield under contracts worth up to $3.2 billion, with a demonstration due by 2028. The list ran from Lockheed Martin, Northrop Grumman, RTX’s Raytheon, General Dynamics and Booz Allen Hamilton to Anduril, True Anomaly, Turion Space, GITAI USA, Quindar and Sci-Tec, with SpaceX one name among twelve. The Congressional Budget Office puts a full interceptor network at up to $542 billion over 20 years.¹² “The most important message I can give to you today is that Golden Dome is real, and it is no longer theoretical,” Gen. Michael Guetlein, the program’s manager, told reporters in remarks reported on April 30. The Pentagon’s own estimate is about $185 billion, with $17 billion directed toward it for fiscal 2027.¹³

The program has already passed its first checkpoint. On August 11, Guetlein told the Army Space and Missile Defense conference in Huntsville, Alabama, that all 12 interceptor contenders appeared to have cleared Gate 1 of a four-gate competition and would move on to Gate 2. “Gate 1 is to get the design in place and do component level testing,” he said, while “Gate 2 is to build a capability that can go into outer space,” and the last two gates prove it works in orbit and inside the wider architecture. Only after all four will the department decide “if we are going to move into production or not.” The 20 contracts behind the $3.2 billion went to those 12 firms.¹⁴

Washington has decided it wants two of everything: two landers, three launch providers, twelve interceptor builders. Every one of those decisions moves money away from a single prime and toward the field.

Washington has decided it wants two of everything: two landers, three launch providers, twelve interceptor builders.

How the Money Flows: What the Suppliers Sell

The money reaches the industry through four kinds of contract. Launch services, paid per mission. Spacecraft, where the customer buys the satellite itself, increasingly bundled with the ride. Components and subsystems, the engines, solar arrays, radios and structures every spacecraft needs regardless of whose name is on it. And connectivity, where the asset in orbit sells a subscription or a government service for years after launch. The best-documented example of a company selling across all four is Rocket Lab, and its last two quarters show how the mechanism works.

Rocket Lab reported second-quarter revenue of $234.1 million on August 10, a record and up 62% from a year earlier, with a backlog of $2.36 billion, up 137%, and more than 90 launches in its launch backlog. It signed more than $437 million of new launch contracts across its Electron, HASTE and Neutron rockets, including options, and a contract to deliver Flatellite spacecraft on Neutron for the Space Force’s SB-AMTI program, where it says it is one of only two vendors delivering launch plus spacecraft. It guided to $250 million to $265 million of third-quarter revenue and said Neutron’s first-stage tank production was aligned with delivering the rocket to the pad in the fourth quarter of 2026. It also lost $49.3 million in the quarter.¹⁵ The SB-AMTI award is worth $397 million.¹⁶

The connectivity leg came on June 29, when Rocket Lab agreed to buy Iridium Communications for about $8 billion in cash and stock, $54 per Iridium share, a 24.1% premium. Iridium operates L-band spectrum and a low Earth orbit network with more than 2.55 million subscribers, many of them government and defense. Rocket Lab shares rose 16% that day and Iridium’s rose 24%.¹⁷ By August 13 the U.S. antitrust waiting period had expired and FCC transfer applications had been filed, with a $3.6 billion bridge loan behind the deal and Iridium shareholder approval still required.¹⁸

✓ Trusted partner See the research behind this story →

The satellite builders are scaling as fast as the launchers. Amazon’s Leo broadband constellation had 396 satellites deployed across 14 launches as of mid-October and has built more than 1,000 in all. Its Kirkland, Washington plant can now make 27 a week, and finished satellites ship to Florida every two or three days. Amazon says it has secured more than 100 launches across multiple providers, with the first Vulcan flight “weeks away” and initial service still planned for 2026, and it is committed to deploying the full constellation by July 30, 2029.¹⁹ Every one of those satellites is a purchase order for components, and every launch is a contract for a rocket that is not Starship.

Put the pieces together and the winners sort into categories rather than a single name. Launch providers with a certified rocket, because the Space Force has shown it will pay a second and a third provider for the same lane. Satellite manufacturers that sell the spacecraft and the ride as one package, because that is how the SB-AMTI contract was written. Component and engine suppliers, because a constellation of 1,000 satellites is a parts bill before it is anything else. And connectivity operators that own spectrum and subscribers, because that is where a launch company goes for revenue that does not depend on the next launch. The forecasts agree: the WEF and McKinsey expect most of the growth to come from industries that use space rather than reach it.²

The Risks: Rockets Explode, Contracts Concentrate, Prices Swing

The risks in this market are not hypothetical. They have all happened this year. On May 28 a New Glenn vehicle exploded during an integrated hot-fire test at Launch Complex 36, destroying the rocket and damaging the pad, and Blue Origin later traced the event to the main oxygen valve on one of its BE-4 engines. Jeff Bezos told Fox News on October 7 that the company was aiming to fly again in December, but on October 16 Reuters reported New Glenn is now expected to return to flight in early 2027, and Blue Origin has not announced a date.²⁰ That one rocket carries NASA’s second lander, an NRO mission and part of Amazon’s constellation.

Schedules slip even without an explosion. Rocket Lab told investors before its August call that the window for a 2026 Neutron debut is “narrowing,” after a first-stage tank rupture during a water-loading test on January 21 pushed the target from mid-2026 to late 2026, and the launch may slide into 2027. Chief executive Peter Beck named the integrated static fire at the pad as the highest-risk milestone, pointing to New Glenn’s accident as the example.¹⁶ A company with a record backlog can still lose money for years while it gets there.

Contracts concentrate. Of the seven Lane 2 missions assigned in October 2025, five went to SpaceX.¹⁰ Starlink alone has about 11,100 satellites in orbit¹, against Amazon’s 396.¹⁹ Golden Dome’s 12 contenders are funding their own prototypes, the program’s blueprint does not schedule actual intercept demonstrations until June 2029, and Guetlein’s office has not “put in any operational requests for production capabilities.”¹⁴ In April he told a House subcommittee, “We are so focused on affordability. If we cannot do it affordably, we will not go into production.”¹²

And prices swing. The largest company in the sector fell from a first-day close of $160.95 to $108 in two months and then rose about 50% in the next two, with billions of insider shares still to unlock.⁵ Rocket Lab moved 16% in a day on a deal announcement.¹⁷ A space stock can reprice on a single test, a single award or a single lockup, and an investor who cannot stomach that should not be in the field.

What to Watch: The Dated Catalysts

The calendar for the next two years is unusually full, and most of the dates belong to companies other than SpaceX.

November 9: Rocket Lab reports third-quarter results after the market close²¹, the quarter it guided to $250 million to $265 million.¹⁵ November: NASA’s briefing timeline puts Starship Flight 15 in that month.⁸ Weeks away, by Amazon’s account: the first Vulcan launch of Leo satellites, with initial service still planned before year-end.¹⁹ December 8: 797.6 million more SpaceX shares unlock after the close, taking the tradable float toward 40%.⁵ December 2026 or early 2027: New Glenn’s return to flight, per the October 7 target and the October 16 report.²⁰ Fourth quarter of 2026: Neutron’s target for reaching the pad¹⁵, with the company itself warning the window is narrowing.¹⁶

Early 2027: SpaceX’s ship-to-ship propellant transfer demonstration. Summer 2027: Artemis III, four astronauts docking with both landers in Earth orbit.⁸ June 2027: Musk’s own SpaceX shares become eligible to trade.⁵ Late 2027: the uncrewed Starship lunar landing⁹, and the window opens for Blue Origin’s NRO launch.¹¹ First quarter of 2028: Blue Origin’s uncrewed Mark 2 landing attempt, then the Artemis IV crewed landing later in 2028.⁹ 2028: the Space Force’s promised initial Golden Dome interceptor capability.¹³ June 2029: the first scheduled intercept demonstrations.¹⁴ July 30, 2029: Amazon’s deadline to have its full constellation deployed.¹⁹

The Market Beyond One Ticker

The second space age is real, and it is arriving faster than the 2024 forecasts expected. But the company that gave it a ticker has said, in two consecutive quarters, that its future is data centers and broadband. Rockets were $1.8 billion of a $14.6 billion quarter and lost money.¹ The customers have responded by refusing to depend on any one supplier. The rivals are reporting record backlogs and buying their way into connectivity. The satellite builders are shipping spacecraft to Florida every two or three days.¹⁹

The question for an investor is no longer whether the space economy will grow. The contracts, the subscribers and the orbital flights have settled that. The question is where the growth lands, and the biggest name in the industry has answered it by pointing its capital somewhere else. The companies that build what the second space age actually needs, the launches, the satellites, the parts and the connections, are not hard to find once you stop looking at only one.