Space, in brief

Blue Origin, Jeff Bezos's rocket company, has moved the next flight of its New Glenn rocket to January 2027. [1] It had aimed for the end of this year. [1]

Its chief executive, Dave Limp, told staff in a memo on Wednesday. [1] He blamed delays rebuilding the launch pad, including bad summer weather. [1]

On Thursday Limp also announced a new engine and a new third stage for the rocket. [3]

New Glenn's delays have slowed Leo, Amazon's internet satellite network. [1] Amazon has turned to other rocket makers meanwhile. [1]

Rockets are short. [9] Quilty Space, a research firm, expects about 150 launches too few by 2030, outside China. [9]

SpaceX's Falcon 9 made 85% of US launches in 2025, and SpaceX plans to wind it down. [8]

NASA has also hired Blue Origin to build a Moon lander. [1]

Rocket engineers call this staging: each stage fires, then separates, and the next one carries on. [6]

GS3 fires after the second stage has gone, so it has less to push. [6] It needs to be light, Limp said, hence its composite tank. [3]

A hiker who drops a heavy bag halfway up climbs the rest more easily.

Every Northstar figure in these reports comes from Blue Origin itself, and the engine has no flight date. [4]

Ars Technica reports Blue Origin aims to launch New Glenn more than 1,000 times a year, eventually. [3] New Glenn has flown 3 times. [4]

A person told CNBC that January is about 10 days past the old target. [1] That figure comes from one unnamed source. [1]

Next comes the January flight, on a booster named Never Tell Me The Odds. [6]

Two big US rockets that customers are waiting for are still on the ground: New Glenn until January and Vulcan until 13 November. [1][7]

A big number such as 1,000 launches a year or 737 satellites sticks in the mind, though the rocket has flown 3 times and the company has its first satellite up. [3][4][10]

Amazon turned to these two rocket makers while New Glenn has been down, and they get the flights. [1]

It has to find other rockets for satellites that were meant to ride on New Glenn. [1]

'You're late,' says the research director at Quilty Space, who has no good answer for them. [9]

Whether January holds for New Glenn.

The memo as reported names a month and no day, and the pad rebuild has already run late. [1]

How much Northstar will really cost to build.

The 25% is what Blue Origin believes it can reach, and it gave no price for either engine. [3]

Who the unnamed customer for GS3 is.

Blue Origin did not name it. [4]

In a test in Texas, Blue Origin's new hydrogen tank, made of light composites, held 150% of its maximum expected pressure for 10 minutes. The company had not been sure such a tank could be built.

Vulcan, the big rocket of US company United Launch Alliance (ULA), is due to fly on 13 November. [7] It last flew 9 months ago. [7]

A solid booster on its side malfunctioned on 2 of its 4 flights. [7] The February launch still succeeded. [7]

After the first fault, in 2024, ULA blamed poor workmanship deep inside a booster's nozzle. [7] A flight in August 2025 went fine, so it seemed solved. [7]

6 months later, in February, the fault came back. [7]

ULA's chief, Mark Peller, told Aviation Week it has found 3 credible causes and acted on all of them. [7]

The investigation is still going on, he said. [7] More nozzle changes could come on later flights. [7]

Blue Origin refitted the valve behind New Glenn's blast on this flight's BE-4 engines. [7] ULA held 30 such engines, most of them already on rockets. [7]

The flight carries 40 satellites for Leo, Amazon's internet network. [7]

Vulcan flights for the US Space Force stay on hold until the results of this flight are in. [7]

It is the first of 38 Vulcan flights for Amazon. [7] ULA has more than 70 missions waiting. [7]

Engineers call this a root-cause hunt: ULA tested materials, parts and a full booster firing to find what broke. [7]

Debris recovered after February's flight gave them real broken parts to study, and that sped things up, Peller said. [7]

A mechanic finds a rattle faster with the broken part in hand.

ULA speaks of 3 credible causes, not 1 proven cause. [7] Peller would not give details of the first fault. [7]

Peller hopes to fly Vulcan once more this year. [7]

Its Vulcan flights wait on hold until the results of an Amazon flight are in. [7]

What really caused Vulcan's booster fault.

The investigation is still open, and Peller says this flight will help show how well the fix works. [7]

CesiumAstro, a Texas maker of satellite radio gear and spacecraft, closed a $730 million funding round on Thursday. [10][11]

$630 million came from selling shares, and $100 million is a loan. [10] That values the company at more than $4.5 billion. [11]

It plans a fleet of 737 satellites called Synchronicity. [10] Its goal is 350 in orbit by 2032. [10]

Until now it sold equipment and spacecraft to others. [11] This fleet would be its own to run. [11]

Its first satellite, Element-1, went up on a SpaceX shared ride in October. [10] It is the first of 8, with launches booked for 2028 and 2029. [10]

It bought 3 companies in the past year, for chips, mission software and AI. [10]

It will spend $500 million over 5 years on its Austin base. [10] It plans to add 550 jobs by 2030. [10]

It is aiming the network at companies, governments and militaries. [11]

Business strategists call this vertical integration: one company owns each step, from chips to satellites. [11]

Founder Shey Sabripour says the firm designs them as one system, so each suits the others. [11] He plans more purchases where a supplier looks at risk. [10]

A baker who grows her own wheat controls every step, and pays for every step.

'Fully successful' is CesiumAstro's own phrase for its first mission, in an announcement printed by SpaceNews. [11]

Why fewer parts make a machine cheaper to build

Every part must be designed, made, checked and fitted to the next

So each part brings work with it, and that work costs money

Manufacturing engineers ask of each one: must this be separate?

This is design for manufacture and assembly, and it merges or removes parts

3D printing helps, because one printed piece can replace several joined ones

With fewer pieces there are fewer steps, so building gets cheaper and faster

A machine rarely gets cheaper because each part gets cheaper. It gets cheaper when some parts stop existing, because their making, checking and fitting go too.

Northstar as a share of today's BE-4, which counts as 100%. These are Blue Origin's own figures, and the engine has not flown.

The panel unrolled with no motor, using energy stored in its own booms, so there was no motor to make, fit or wire up.

It designs the chip, antenna and spacecraft as one system, so fewer pieces arrive from outside firms and then have to be made to fit.

Orca fits its software once to the flight software Orbital uses, so that work is not repeated for each satellite in the fleet.

Merging or sharing parts has a price too.

An engineer deciding to remove a part sets the price a satellite company pays for a ride, and that satellite may carry someone's phone signal.

CesiumAstro plans 550 more jobs there by 2030, as part of a $500 million expansion. [10]

How CesiumAstro gets hundreds of satellites to orbit.

Its chief says it is still working out which partnerships it needs. [10]

NASA announced its next experimental plane, the X-70 Fractal, on Wednesday 4 November. [12]

X-planes are test craft for new flying ideas. [13] The Bell X-1, which first broke the sound barrier, was one. [13]

The X-70 will carry a crew and cruise about as fast as a Boeing 737. [12][13] It can reach 45,000 feet. [12]

It can carry about 900 kg of test equipment, inside or outside. [12]

NASA chief Jared Isaacman expects a first flight within 2 years. [12]

Its most recent X-plane, the X-59, has flown for over a year. [12] It is built to pass the sound barrier without a loud boom. [12]

NASA wants to test how a plane can turn and move in ways 'never before demonstrated in crewed flight'. [12]

Aircraft engineers call this a test bed: one plane that carries many experiments. [12]

Isaacman says its equipment can be swapped quickly, so teams move fast from one experiment to the next. [13]

A school science room keeps one bench and changes the experiment each week.

NASA has shown only an AI-made picture so far, and Space.com says it reveals little. [12]

Space.com notes a similar outline and weight to the Model 437 Vanguard, a Northrop Grumman test plane. [12] Neither report says who is building the X-70. [12][13]

Deutsche Telekom, which owns T-Mobile, says SpaceX needs at least $200 billion to become a big US phone company. [15]

Its chief, Tim Hottges, said so to investors on Thursday. [15]

SpaceX's plan, Starlink Mobile, joins satellites that talk directly to phones with a network on the ground. [15]

Hottges compared spectrum, the radio bands phones use. [15] SpaceX has bought 76 megahertz, and his company has 400. [15]

Catching up would cost three-figure billions, he said. [15] Renting towers would be another $20 billion to $30 billion. [15]

Hottges still takes Musk 'super serious'. [15] He said the car industry ignored the threat for too long. [15]

Economists call this a barrier to entry: what a newcomer must spend before it has its first customer. [15]

T-Mobile's owner already holds 400 megahertz and 85,000 towers. [15] SpaceX must buy or rent the same things first. [15]

A new pizza shop pays for ovens and a lease before it sells one slice.

The $200 billion is a rival's own sum, given to its own investors. [15]

It already holds spectrum and towers that SpaceX would have to buy or rent first. [15]

How much SpaceX would truly spend to build a phone network.

A recent report says it could tap 80,000 US towers, and the article gives no figure from SpaceX. [15]

Orca Aerospace, a start-up, has a deal worth up to $20 million with Orbital, a Los Angeles company. [16]

Orbital plans a fleet of data-centre satellites: 10 in 2029, 100 in 2030 and 1,000 in 2031. [16]

Orca's software would steer each satellite and handle faults, asking people only when needed. [16]

A small AI model suggests moves. [16] A safety layer checks each command against Orbital's rules before it is sent. [16]

It first runs in 'shadow mode', sending its data down for people to review. [16]

It is due to fly on Orbital-1 in 2028. [16]

Orbital's chief, Euwyn Poon, says nobody runs a central map of these paths. [16] So each satellite must steer for itself. [16]

Many data-centre fleets want the same orbits, whose paths meet over the poles. [16] That will likely mean many dodges to avoid crashes. [16]

Orca's chief, Jacob Ososke, calls this gaining autonomy in layers: a program takes on jobs one step at a time. [16]

Orca's software starts by reporting what it would do, and people check that. [16] It takes on more only in steps. [16]

A learner driver starts with an instructor beside them, ready to brake.

This is a letter of intent. [16] Orbital's Pathfinder test craft is planned for July 2027, and its fleet starts in 2029. [16]

Orca fits its software once to the flight software Orbital uses, so that work is not repeated for each satellite in the fleet.

It earns a licence fee for each satellite running its software, so a bigger Orbital fleet pays it more. [16]

Orbital's chief says Orca's software replaces the need for a team of them on his fleet. [16]

A rock about 2 metres wide, called 2026 QQ8, is circling Earth for a while. [14]

It was caught in April, first seen in August and reported on 29 October. [14]

Its closest point is about 568,000 km away. [14] That is about 177,000 km beyond the Moon's average distance. [14]

It goes round the Sun in 359 days, and it nears Earth about once every 10 years. [14]

It stays through December and breaks away in 2027. [14]

If it ever hit Earth, it would likely be destroyed in the air first, being so small. [14]

Research last year suggested Earth could hold as many as 6 at once. [14] Many may be broken bits of our own Moon. [14]

Astronomers call this a temporary satellite: a rock held by Earth's gravity for a short time. [14]

Its year, 0.98 of Earth's, is almost the same as ours. [14] This time it came so close that Earth's gravity could seize it. [14]

A friend jogging beside you is easy to catch by the sleeve, and one sprinting past is not.

Space.com's January 2027 date is worked out from a model's average stay of 9 months. [14]

Nathan Moore of Sandia National Laboratories, a US research lab, led a simulation of pushing asteroids with a nuclear blast. [17]

The study, in the journal Icarus, used 3 very differently shaped asteroids: Bennu, Itokawa and Golevka. [17]

It is a last resort, for a rock found too late for a simple crash like NASA's DART test. [17]

The blast goes off at a distance. [17] X-rays boil one side of the surface, and that makes most of the push. [17]

Which side faced the blast could change the push up to nearly 5 times on Golevka. [17] It was nearly 4 times on Itokawa and 2 times on Bennu. [17]

Its new program runs over 1,600 times faster, Moore said. [17] Older ones could not handle real shapes. [17]

A rock heading our way may have an unknown shape and show any side. [17] So the authors say it could be mapped from a spacecraft before any blast. [17]

Physicists call this recoil: vapour flying off one side pushes the rock the other way. [17]

A lumpy rock turns a different face to the blast from each side. [17] So the same blast gives a different push. [17]

Kick a football and it flies straight, but a rugby ball's path depends on where your foot lands.

This is a computer model. [17] It covers rocks breaking on the surface, Moore says, not the whole asteroid. [17]

A new computer program at Sandia, a US research lab, runs over 1,600 times faster than older ones. It can test how to push real, lumpy asteroids that the old models could not handle.

The James Webb telescope has found many bright galaxies from a few hundred million years after the Big Bang. [18]

Their stars sit in very dense clusters, and why those galaxies look so bright is a puzzle. [18]

Shyam Menon of the Flatiron Institute and Aaron Smith of the University of Texas at Dallas offer a possible answer. [18]

It is Lyman-alpha, a kind of light given off by hydrogen, the most common element in the universe. [18]

Thick gas forming stars stays put, because its gravity is too strong for this light. [18] Thinner gas and dust get blown away. [18]

So stars keep forming, and the cleared gas lets their light out, the study in The Astrophysical Journal suggests. [18]

Menon says it can clear the 'fog', so we see early stars more clearly. [18]

Astronomers call this radiation pressure: light pushes on whatever it hits. [18]

Hydrogen soaks up this light and throws it back out almost at once, again and again. [18] That cycle lets it carry a lot of force. [18]

A ball bouncing round a small room keeps hitting the walls, and each hit is a small shove.

The result comes from sums on snapshots of earlier computer simulations, not from watching a galaxy do it. [18]

The Vera Rubin Observatory, a telescope on a mountain in Chile, has found 2 faint companions of the Milky Way. [19]

Aquarius IV is most likely a tiny dim galaxy. [19] Rubin-GC1 is most likely a globular cluster, a tight ball of stars. [19]

Both came from a preview of its data, released in July 2026. [19]

Rubin-GC1 lies toward the Milky Way's crowded, dusty middle, which makes it harder to find. [19] Its stars are about 9.2 billion years old. [19]

Astronomers already knew about 65 small galaxies and about 40 globular clusters around the Milky Way. [19]

These small companions are pieces of how the Milky Way grew, by swallowing smaller galaxies. [19]

They also help show where dark matter, the unseen stuff in galaxies, gathers in clumps. [19]

Astronomers call this a satellite census: counting every small companion of a galaxy. [19]

They use a filter map that hides stars of the wrong colour and brightness. [19] A faint group then stands out. [19]

To find your class in a packed stadium, you look only for the school colour.

Aquarius IV is not settled. [19] Few of its stars are measured so far, and it could still be a cluster. [19]

The Vera Rubin Observatory found 2 faint companions of the Milky Way using only a preview of its data. Its full survey runs for 10 years.

Hours into Thursday's spacewalk, controllers saw the space station's last new solar panel giving about half the current expected. [20]

'We are not necessarily going to deploy,' Mission Control told the astronauts. [21]

They were Anil Menon of NASA and Joshua Kutryk of the Canadian Space Agency, fitting the last of 8 panels. [21]

Kutryk unplugged one connection in the next dark spell, checked both ends and plugged it back in. [20]

The current came back good. [20] The panel then unrolled by itself in about 10 minutes, with no motor. [21]

The panels raise the station's power by up to 30%. [20] It is due to retire in late 2030. [21]

Electricians call this isolating: making a circuit dead before anyone touches it. [20]

The station's wings and cables are live in sunlight. [20] So cables can be handled only in Earth's shadow, about 45 minutes of each 90-minute orbit. [20]

You switch off the power at the wall before changing a light fitting.

The 8 panels are now all up, finishing work begun in June 2021. [20]

The next US spacewalks are aimed at early January, to upgrade a cosmic-ray detector. [20]