Space, in brief

NASA's next crew flight to the space station will not launch on 12 September. An oxidiser leak was found in the propulsion system of its SpaceX Dragon capsule. [1]

NASA announced the delay on 29 August. It has not set a new date. [1]

Four people are waiting on it: Jessica Watkins and Luke Delaney of NASA, Josh Kutryk of the Canadian Space Agency, and Sergey Teteryatnikov of Roscosmos, Russia's space corporation. Their stay is meant to last seven months. [1]

There is no second vehicle to switch to. Boeing's Starliner capsule, which NASA funded so it would never depend on one company, has not carried a crew since 2024. [2]

That 2024 flight went wrong. Its two astronauts came home on a SpaceX Dragon instead. [2]

The rocket underneath Dragon is being wound down by its own maker. SpaceX plans to retire the Falcon 9 in favour of Starship, the much bigger rocket it is still testing. [2][3]

The wind-down has started. A Falcon 9 flight in late August was the last planned Starlink launch from Florida, after about 260 of them; those missions move to Starship. [6]

Falcon 9 is not failing. This week a booster called 1063 flew for the 35th time, two flights short of SpaceX's own reuse record. [5]

Phil McAlister, who worked at NASA for almost 20 years until he retired last year, wrote that the agency is about to lose its way to low Earth orbit. [2]

NASA's administrator, Jared Isaacman, answered him in public late on Wednesday night. You cannot simply slap the word commercial on every problem, he wrote. [2]

Isaacman listed what NASA is already trying to do at once: return to the Moon, build a base there, get nuclear power under way, fly more science missions, and replace the space station with a commercial service. [2]

Seven people are living on the station now. A Russian Progress cargo ship undocks on 7 September and its replacement launches from Kazakhstan on 9 September. [4]

Why it matters. Watch the new crew date against Starship's next test, which SpaceX is thought to be aiming at mid-September. If the crew slips past it, the rocket meant to replace Falcon 9 will have reached orbit first. [1]

India's space agency ISRO set a GSLV Mark II rocket to send EOS-05, a 2,367-kilogram Earth-imaging satellite, toward geosynchronous orbit at 5:25 p.m. Eastern time on 3 September. It would be India's first imaging satellite at that height. [7]

Why it matters. At 35,786 km up a satellite keeps pace with Earth's turn and hovers over one patch of ground, so it can watch continuously instead of passing over twice a day. India lost its last attempt when a GSLV third stage failed in 2021. [7]

A Falcon 9 launch in late August was the last planned Starlink mission from Cape Canaveral. From here Starlink flies from Florida on Starship, SpaceX's launch vice-president Kiko Dontchev said; about 260 Starlink missions had gone up from there on Falcon 9. [6]

Why it matters. It is the clearest sign yet that the rocket NASA's crews depend on is being retired by the company that owns it.

A new national space transportation policy signed by US President Donald Trump calls for more spaceports, to support more than 1,000 launches and reentries a year by 2030. It tells the US defence department and NASA to coordinate on ground infrastructure and to publish range schedules. [6]

Why it matters. Launch pads and range slots, not rockets, are what the industry says it will run out of first. Most launch schedules are not public until days or weeks beforehand. [6]

NASA restarted two of the three instruments on its 21-year-old Swift observatory on 26 August. They had been dark since February to cut the spacecraft's drag through the thin upper atmosphere and keep it in orbit long enough to be rescued. [8][9]

Why it matters. The rescue was abandoned. A robot craft called LINK, launched on 3 July, lost two of its three reaction wheels and part of its thruster system, and its mission was cut back to a rendezvous demonstration. [9]

Spain's PLD Space closed a 108 million euro extension round led by Mitsubishi Electric, and Germany's HyImpulse closed over 50 million euros. PLD has now raised 488 million euros since 2011 and wants 30 launches a year by 2030. [10]

Why it matters. European launch money used to lag well behind the United States. Both companies are still short of orbit: PLD aims for a first orbital test in late 2026, HyImpulse in late 2027 or early 2028. [10]

Two launch integrators, Exolaunch and SEOPS, have bought Falcon 9 flights to run shared-ride programmes of their own. A French broker called RIDE! signed with MaiaSpace for spare room on up to six flights from 2028. [3]

Why it matters. Customers stop being customers when they think the service may end. The fear at the Small Satellite Conference in Utah was that SpaceX shared rides would disappear the way Northrop Grumman's Pegasus rocket did. [3]

At the same conference, Luxembourg's Edge Aerospace hired Redwire to build its data processing units in the United States. The ground-station operator KSAT agreed with Unibap to put computing on two demonstration satellites, launching no earlier than late 2027. [3]

Why it matters. Sending every raw image down to the ground is the slow, expensive part. Doing the sums in orbit and sending the answer is the way around a crowded downlink. [3]

General Douglas Schiess became the third Chief of Space Operations of the US Space Force at a ceremony at Joint Base Andrews in Maryland on 3 September, succeeding General Chance Saltzman. [11]

Why it matters. He takes over a service whose subject has changed. Militaries now treat satellites as weapons systems, and a Chinese spaceplane released a small satellite in June that looped around another Chinese satellite and came back. [18]

The US Space Force's latest round of small-business research contracts favours technology near operational use rather than open-ended research, a shift from a programme that had drawn scepticism from the industry it funds. [12]

Why it matters. Research money that never reaches a working system is the complaint small suppliers make most often about defence contracts.

Sarmony, a Japanese radar-imaging startup, signed on 25 August to launch its first demonstration satellite on Space One's Kairos rocket in 2028. Kairos has flown three times and failed to reach orbit each time, breaking apart just over a minute after liftoff in March. [6]

Why it matters. When launch is scarce, buyers take rides on vehicles with no successful flights behind them. Sarmony's satellite uses the flat DiskSat shape, which the US research centre Aerospace Corporation demonstrated with four craft launched eight months ago and is now pushing at the industry. [19]

Researchers used Sentinel-1 radar satellites to watch three emperor penguin colonies through the Antarctic winter from 2017 to 2024. All three moved, and groups within them sometimes shifted away from the ice edge together even while it looked stable. [13]

Why it matters. Emperor penguins are the only bird that breeds through the polar night, which is exactly when ordinary camera satellites see nothing. Radar makes its own light, so it works in the dark. [13]

A group of theoretical physicists proposed that black holes formed in the early universe before any stars did, using heat from the cosmic microwave background, the faint glow left over from the Big Bang. [14]

Why it matters. It is an attempt to explain the little red dots the James Webb telescope found: distant galaxies that appear to hold surprising numbers of astonishingly ancient black holes. [14]

A study led by Christian Renggli of the Max Planck Institute for Solar System Research used the silica content read from infrared light to argue that Mercury was shaped by more powerful volcanism than thought. Its lava plains are over a kilometre thick and date from 4.1 to 3.5 billion years ago. [15]

Why it matters. Silica controls how thick and sticky magma is, so reading it from orbit tells you what kind of eruptions built a surface you cannot visit. [15]

Comet 220P/McNaught, which loops around the Sun every 5.5 years, had two outbursts after passing closest to the Sun on 14 June that took it to 20,000 times its usual brightness. It is now at magnitude 9.2 and fading, still within reach of a small telescope. [16]

Why it matters. Sunlight turns a comet's ices straight to gas, and the dust that comes with it reflects the light. Outbursts are not predictable, so the show arrived without warning. [16]

A study in Nature trained a machine-learning model on debris population, mass and near-miss data across the range of orbital heights and tilts. It found that the number of satellites can rise a lot for only a moderate rise in predicted collision risk. [17]

Why it matters. Risk is not spread evenly through orbit, so where a fleet is placed matters as much as how many satellites it holds. [17]

Roscosmos flight engineer Pyotr Dubrov is loading rubbish and old equipment into the Progress 94 cargo craft before it undocks on 7 September. Progress 96 launches from Baikonur in Kazakhstan on 9 September and docks two days later. [4]

Why it matters. Cargo keeps arriving on schedule while the crew rotation waits, which is why nobody on board is in any danger from the delay.

Space writer Jatan Mehta argued in a six-part series that the recent success of Skyroot's Vikram-1 rocket does not close India's shortage of orbital launch capacity, because the country is short of spaceports, suppliers and medium and heavy rockets. [6]

Why it matters. A small rocket working is not the same as a country being able to launch what it wants, which is the same gap Europe is now spending to close.