SpaceX's Starship refuelling math is still unresolved as Artemis IV planning continues

Space.com published an analysis on 3 November that lays out the problem plainly. NASA has not settled how many Starship tanker launches Artemis IV will need. The refuelling sequence itself remains a graphic on a slide, not a flown procedure. The piece speaks directly to how a space agency approves and pays for a mission whose architecture is still moving.

The spread of numbers is the story. A Government Accountability Office audit from summer 2021 put the requirement at up to 16 launches. Earlier this month, a member of the Aerospace Safety Advisory Panel said it could be as high as 22. SpaceX founder and CEO Elon Musk responded on X that it would be closer to six or eight, according to Space.com. No one has published a validated figure since.

That gap matters because Space.com describes Artemis IV as the programme's first crewed lunar landing. The published order of operations has an SLS rocket carrying four astronauts in an Orion capsule to high Earth orbit. There they dock with the Starship lander. The two spacecraft then burn for the Moon and separate in low lunar orbit. Two astronauts stay aboard Orion while the other two descend. Everything upstream of that descent depends on propellant launched separately, stored in orbit and transferred. Space.com notes that NASA released a graphic during a recent Artemis update outlining the sequence. But a graphic is not a flight plan. The number of tanker flights drives cost, schedule and the agency's own risk assessments.

Approvals are landing elsewhere

The same week produced a cleaner example of a space approval that did resolve. That is the UK Space Agency's Active Debris Removal Mission, which the agency's own blog describes as the UK's first national space mission. The blog, dated 31 October, says ClearSpace UK will build the spacecraft. It will grab two defunct satellites and guide them back towards Earth to burn up in the atmosphere. The agency is running a public competition to name the mission, closing on Thursday 10 December 2026.

The blog supplies the numbers the agency wants the public to hold in mind. Nearly 13,000 working satellites are in orbit today, and by 2030 there could be around 60,000. It also states that debris travels at roughly 10 times the speed of a bullet. Those are the agency's figures, not an independent audit. The blog does not give a launch date or a mission cost.

On the commercial side, the approval that actually closed this week was a shareholder vote. The Robot Report reported on 3 November that more than 84% of votes cast at a special meeting on 2 November backed Churchill Capital Corp XI's combination with Agility Robotics. The deal values the humanoid developer at a $2.5 billion pre-money equity value. The SPAC is expected to close on 4 November, according to the same report. Trading on Nasdaq under the ticker AGLT is expected to begin on 5 November. The money shrank on the way through. The Robot Report says the deal is now expected to provide approximately $465 million in gross capital before expenses. That is down from more than $620 million anticipated when the merger was announced in June. The revised figure comprises roughly $164 million from Churchill's trust account, about $201 million from a PIPE led by Foxconn, and $100 million in equity Agility received after the initial announcement.

Agility's orders are the part worth reading carefully. The company has claimed more than $300 million in committed multi-year orders for Digit 5, its next-generation humanoid unveiled in September. The Robot Report flags that this is not revenue. The multi-year value is expected to be realised over time and remains subject to contractual milestones, product features and specifications. Digit robots are already operating in 9 customer environments, including a GXO Logistics facility in Flowery Branch, Georgia. Expansions were announced in October to Statesville, North Carolina, and Lebanon, Tennessee.

Hardware with a deadline

Then there is the approval nobody granted. The Verge reported on 3 November that NASA's Swift Observatory is expected to re-enter Earth's atmosphere on Wednesday at approximately 3:30AM ET, give or take 11 hours. That is almost 22 years after launch. NASA gave up on a rescue earlier this year, according to The Verge. The Link spacecraft attempting to raise its orbit ran into attitude control problems.

The observatory weighs around 3,200 pounds and measures 5.6 metres high and 5.4 metres wide. Much of the latter is deployable solar arrays. NASA's stated risk of harm to anyone on Earth is approximately 1 in 2,100. Surviving fragments are limited to between 20.6 degrees south and 20.6 degrees north latitude because of Swift's low inclination. The agency said it and Space Force will keep monitoring the re-entry and updating predictions.

Regulatory friction, meanwhile, is landing on SpaceX from a different direction. Tom's Hardware reported on 2 November that a judge refused to dismiss a lawsuit over pollution from xAI data centre gas turbines in Mississippi. The judge ruled the turbines are not exempt from the Clean Air Act. CNBC reported on 31 October that SpaceX also lost a bid to toss an NAACP lawsuit over air pollution at a Memphis data centre. Neither ruling settles the underlying claims, but both keep them alive.

The company's terrestrial ambitions have their own approval problem. Light Reading reported on 3 November that analysts have poured cold water on the idea that acquiring Boost Mobile would let SpaceX inherit EchoStar's MVNO agreements with AT&T and T-Mobile. The report followed a Washington Post story that Musk and EchoStar's Charlie Ergen had discussed a deal. LightShed Partners analysts Walt Piecyk and Joe Galone said in a research note that SpaceX "cannot simply acquire Boost Mobile and inherit its MVNO agreements with AT&T and T-Mobile". They pointed to change-of-control provisions.

"Every MVNO contract has a change-of-control provision," a person familiar with the MVNO business told Light Reading.

The filings back that up. Light Reading cites EchoStar's SEC filings on the ten-year network services agreement struck via its spectrum sale to AT&T. Those filings restrict resale to third parties including "certain large technology companies". AT&T retains the right to terminate and a two-year transition period. MoffettNathanson analyst Craig Moffett noted that Boost's MVNO deal with T-Mobile expires on 1 July 2027. He said that makes it effectively irrelevant. New Street Research analyst David Barden called the Post's suggestion that SpaceX "may need consent" from the two carriers "a gross understatement", according to Light Reading.

What the record shows

Set these side by side and a pattern appears in how space approvals actually work in late 2026. A civil agency can name a mission and open a competition for its name while leaving cost and launch date unstated, as the UK Space Agency has done. A commercial venture can clear a shareholder vote and still see its expected proceeds fall between announcement and closing. A regulator can decline to dismiss a case without deciding it. And the largest technical question in the Artemis chain, the number of refuelling flights, sits with no agreed answer.

Space.com's account sets the 2021 GAO estimate of up to 16, the safety panel's 22, and Musk's six to eight against each other without resolving them. That is the honest state of the record. No NASA figure supersedes any of the three. Older reporting fills in how the agency got here. Ars Technica argued on 30 October that SpaceX no longer needs NASA and that this is a problem for the American space agency. That piece reads differently next to a refuelling plan that NASA has to certify for its own crewed landing. SpaceX's Starfall capsule, reported by Space.com on 29 October, delivered Air Force Research Laboratory cargo pods for a point-to-point delivery test. It launched on a Falcon 9 rideshare mission on 20 October and splashed down in the Pacific. The Air Force Research Laboratory said the pods were recovered with SpaceX support in an X post on 26 October.

None of that resolves the tanker count. It does show a company accumulating military and commercial work that is not dependent on a lunar lander flying on schedule.

For the agencies, the approval calendar is crowded and uneven. Rocket Lab launched a Japanese Earth-observing satellite on its 99th mission to date, Space.com reported on 2 November. NASA's Artemis II is the next crewed flight in the queue, and the Starship refuelling question belongs to the mission after it. The UK's debris removal spacecraft has a name competition but no published launch date. Swift has a re-entry time and a 1 in 2,100 risk figure.

Only one of those has a number that everyone involved accepts. The rest are estimates. At least one of them, the refuelling count, has been publicly disputed by the company that has to fly it.