Daylila

Space · Wednesday, 19 August 2026

01 · Briefing · what happened

A vest flown around the Moon cut a solar-storm dose by 60%

Space 7 min 23 sources

Three findings this week were about the environment itself. A first flight test of wearable radiation shielding. A re-reading of storm data that removes the ceiling we thought protected us. And a defence programme quietly buying up firms whose chips survive being hit. Meanwhile two rockets came home: a Chinese booster landed on its legs in the Gobi, and SpaceX towed a Starship into port after 24 days at sea.

60%

dose cut by the vest

against a 1972-class solar storm

2x

worse extreme storms

once the measurement bias is removed

390 km

downrange landing

LandSpace booster, Minqin County, Gansu

$71.3bn

proposed Space Force budget

a 120% rise on 2026

At a glance

  • A vest called AstroRad flew around the Moon on Artemis I, fitted to one of two detector-packed mannequins.
  • Against a 1972-class solar storm it would have cut the dose by about 60%, from 222 millisieverts to 87.5.
  • That matches the Orion capsule's built-in storm shelter, but you can keep working while wearing it.
  • A Nature paper says the apparent limit on how hard storms hit Earth was a measurement artefact, not physics.
  • Corrected, the worst storms look about twice as damaging to grids and satellites as models assumed.
  • Machines get the opposite treatment from bodies: not shielding, but chips built to survive being hit.
  • China's LandSpace landed an orbital booster on its legs in the Gobi, a first for a private Chinese firm.
  • SpaceX towed an intact Starship into port after 24 days at sea, its first real look at a flown heat shield.

Forces in play

Radiation risk Building

A Nature reanalysis by NASA Goddard removes the ceiling scientists assumed protected Earth. A rare severe solar storm could hit grids and satellites about twice as hard as models say.

Countermeasures Easing

The AstroRad vest flown on Artemis I cut a simulated 1972-storm dose by 60%, the first flight evidence that a wearable shield works beyond Earth's magnetic field.

Reusable rockets Building

LandSpace landed a Zhuque-3 booster on its legs in Gansu and SpaceX recovered an intact Starship, so two more vehicles are moving toward flying again rather than being thrown away.

Launch shortage High

About 270 orbital launches a year, and satellite firms still cannot buy a ride. Quilty Space's Caleb Henry calls it an industry in panic; Amazon has slowed satellite production for want of launches.

Lunar debris Building

A four-tonne Falcon 9 stage punched a fresh crater near Einstein Crater at 5,400 mph, joining roughly 3,000 objects on a surface with no disposal rules at all.

Military money High

The proposed 2027 Space Force budget of $71.3 billion is the biggest percentage jump for any US service since 1952. Golden Dome contractors are already buying firms whose chips survive being hit by particles.

In play NASA and DLR — flew the AstroRad experiment on Artemis I and published the first results Nithin Sivadas, NASA Goddard — led the Nature paper removing the assumed storm ceiling LandSpace — first private Chinese company to land an orbital booster SpaceX — recovered an intact Starship and flew its 100th mission of 2026 Voyager Technologies — bought radiation-hardened electronics firms ahead of Golden Dome

How it unfolded

  1. 2022 Artemis I carries two mannequins, one wearing the AstroRad vest
  2. 5 Aug a spent Falcon 9 stage crashes into the Moon near Einstein Crater
  3. 13 Aug Science Advances publishes the first wearable-shielding flight results
  4. 15 Aug SpaceX launches two Falcon 9s 38 minutes apart, a new record
  5. 18 Aug LandSpace lands a booster in Gansu; a Starship reaches port after 24 days

Where this points

The next test is whether LandSpace actually reflies the booster it just recovered. It aimed at a late-2026 reuse earlier in the year but gave no date after this flight, and reuse is what would start easing the launch shortage.

Full briefing

The vest that worked

When NASA flew Artemis I around the Moon in 2022, it carried two headless torso mannequins fitted with radiation detectors. One, named Zohar, wore a vest called AstroRad. The other, Helga, wore nothing. The results were published this week in Science Advances, and they are the first flight data anyone has on wearable radiation shielding beyond Earth’s magnetic field [1].

The vest is flexible, with a core of high-density polyethylene shaped to sit over bone marrow and the organs most easily damaged. Run the measured numbers against real solar storms and the vest performs well. In a 1972-class event it would have cut a crew member’s dose by about 60%, from 222 millisieverts to 87.5. In a repeat of 1989 the cut would be roughly 40% [1].

That is about the same protection as the storm shelter already built into the Orion capsule. The difference is what you can do while protected. A shelter is a cupboard: you sit in it and wait. A vest you can wear while fixing something [1].

And the storms may have no ceiling

For decades, space physicists have noticed something reassuring. Push the solar wind harder and Earth’s magnetic response rises with it - but only up to a point. Past that, the response seemed to flatten out, as if the planet had a built-in limit.

A Nature paper published in July, and written up this week, argues that the flattening was never real. It comes from Nithin Sivadas of NASA’s Goddard Space Flight Center and Maria Walach of Lancaster University [2]. The limit, they say, is an artefact of where we measure from.

Nearly all solar-wind readings come from spacecraft parked at a point called L1, about 1.5 million km closer to the Sun than we are. The wind changes on the way in, and timing is uncertain. So the most extreme reading at L1 is usually paired with a less extreme event at Earth, purely by chance. That is a well-known statistical trap called regression to the mean: measure something noisily, and the biggest measurements are the ones most likely to be flukes [2].

Correct for it, and the flattening vanishes. The relationship stays straight all the way up. In plain terms, a one-in-a-thousand-year storm could hit power grids and satellite fleets about twice as hard as the models say [2].

The machines get a different answer

Bodies get shielding. Machines mostly do not, because shielding is heavy and a charged particle that hits a chip does not need to deliver much energy to do damage. It only has to land in the wrong place.

That is why “radiation-hardened electronics” showed up this week in an unlikely place: a Scientific American account of how Golden Dome, the US missile-defence programme, might actually be built. Voyager Technologies, one of the firms angling for the work, spent the last two years buying companies that already make space hardware. Among them were makers of radiation-hardened electronics, electric thrusters and solid-fuel motors [3]. Nobody buys a chip company for fun. They bought it because putting weapons in orbit means putting computers in orbit, and orbital computers are a specialist trade.

The scale explains the appetite. One analysis, by Todd Harrison of the American Enterprise Institute, puts the requirement at roughly 1,900 interceptors. That is what it takes to keep continuous cover over every point on Earth with two shots at each threat [3]. There is no published design yet. As Harrison puts it, “There is no architecture” [3].

The money is real even where the architecture is not. A new Aerospace Corporation analysis of the proposed 2027 budget puts the Space Force at $71.3 billion. That is a 120% rise on 2026, and the largest percentage increase for any US military service since 1952. The growth sits in missile warning, moving-target tracking, command and control, and classified programmes [4].

Two rockets came home

On 18 August the Beijing company LandSpace launched its second Zhuque-3 rocket from Jiuquan. The first stage flew back down and landed on its legs about 390 km downrange, in Minqin County, Gansu [5]. It is the second Chinese orbital booster recovery ever, after a state-owned Long March 10B on 10 July. It is the first by a private Chinese company, and the first Chinese recovery on land using landing legs [6].

The first Zhuque-3 flight, eight and a half months earlier, reached orbit but crashed its booster during the landing burn. LandSpace says it fixed that by using fewer engines in the landing burn, adding a predicted landing point to the onboard control system, and improving the heat protection [5]. The rocket lifts 21,000 kg to low orbit expendable, or 18,300 kg when the stage comes back - roughly Falcon 9 territory [5].

The same week, SpaceX finally got a Starship back. The vehicle from July’s thirteenth test flight splashed down in the Indian Ocean on 24 July and, unusually, did not break up. A recovery ship towed it engine-first for 24 days and reached Christmas Island on 18 August [7]. That matters because the heat shield, some 18,000 ceramic tiles, is the company’s central unsolved problem. Until now every ship has been judged only through sensors and camera footage [7].

Volume kept climbing regardless. SpaceX flew its 100th mission of 2026 on 18 August, its third straight year past the century mark [8]. On 15 August it put two Falcon 9s up 38 minutes apart from opposite coasts, beating its own record by 27 minutes [9]. Blue Origin is adding a second pad at Cape Canaveral while repairing the first after a New Glenn exploded there in May [10]. China returned to flight on 16 August with back-to-back launches, six days after a Long March 7A broke up in flight [11].

None of that has fixed the shortage. Satellite operators still cannot buy rides: about 270 orbital launches a year, prices never lower, and Quilty Space’s research director Caleb Henry describes “an industry in panic” [12].

The Moon is filling up with junk

A spent Falcon 9 upper stage weighing about four tonnes hit the Moon near Einstein Crater on 5 August at 5,400 mph [13][14]. South Korea’s Danuri orbiter and NASA’s Lunar Reconnaissance Orbiter both photographed the new crater; a Chinese commercial debris-tracking satellite recorded the impact [14]. Danuri’s team had spotted the incoming stage in late June and could not rule out a close pass. A manoeuvre planned for an eclipse happened to move the orbiter clear [14].

Around 3,000 human-made objects and roughly 200 tonnes of material now sit on the lunar surface [13]. Almost none of it is governed. The only international rules that touch lunar disposal are the COSPAR planetary-protection guidelines, written to stop biological contamination of scientifically interesting sites, not to manage where spent rockets fall [15]. Crewed landings are two to four years away [15].

“We’re in a transitional moment,” the astronomer and space historian Jonathan McDowell told Space.com. Governance, he said, has not kept up [13].

Also this week

Astronomers using Webb found three active supermassive black holes inside one galaxy 12.5 billion light-years away. We see it as it was less than 1.3 billion years after the Big Bang. Lead author Hannah Ubler of the Max Planck Institute for Extraterrestrial Physics calls it the first such trio found in the distant universe [16]. Curiosity, marking 14 years on Mars, photographed its first field of polygon cracks, a honeycomb pattern thought to form from repeated wetting and drying [17].

NASA administrator Jared Isaacman said he is “extremely confident” Artemis 3 will fly in 2027, with SLS stacking under way at Kennedy [18]. Mike Fincke retired after exactly 30 years, four spaceflights, two ISS commands, and the rare distinction of having launched on three different spacecraft [19]. ESA’s Sophie Adenot became the first French woman to walk in space, replacing a station antenna with NASA’s Anil Menon [21]. And NASA’s early-concepts programme funded SPARK, a swarm of small spherical flying robots meant one day to explore the caves of Saturn’s moon Titan [20].

Underneath all of it sits a question nobody has settled. Ars Technica argued this week that the United States has not yet reckoned with what it would mean for China to land people on the Moon first. Nor with what China might claim where its rovers drive [22]. The Artemis II crew, back from lunar orbit in April, spent the week saying thank you [23].

02 · Lesson · why it matters

Why the computer flying a spacecraft is older than your phone

Outside Earth's magnetic shield a single particle can flip a stored bit, so the safe design is bigger, slower and deliberately decades behind.

How it works

  1. Outside Earth's magnetic field, charged particles arrive constantly
  2. One particle can flip a stored bit or short a circuit
  3. A bigger, slower transistor holds more charge, so one hit matters less
  4. So flight computers are deliberately old, huge and expensive
  5. Nearer Earth the field still shields you, so cheap parts plus spares work
  6. The right answer depends entirely on where the machine is going

The twist

The computer flying a billion-dollar spacecraft is decades behind your phone on purpose. Where one particle can flip a stored bit, bigger and slower is the safer engineering, not the lazier.

Where you've seen this

Hospital equipment

an infusion pump runs proven old chips because a reboot is not an option

Railway signalling

signal boxes keep decades-old logic because it fails in ways people understand

Bank ledgers

core accounting still runs mainframe code nobody dares replace mid-flight

Aircraft avionics

flight controls are certified once and frozen, not updated on a phone cycle

The catch

Old and slow is the right answer for deep space, not everywhere. In low orbit the planet still does most of the shielding, so cheap fast parts plus spare copies increasingly win.

Full lesson

Two answers to the same problem

This week gave the same hazard two completely different engineering answers.

For a human body, the answer was a vest: a slab of plastic shaped to sit over the parts that damage easiest, cutting a solar-storm dose by about 60%. Put material between the particles and the person.

For a machine, the answer was a company purchase. A defence contractor stocking up for orbital work bought firms that make radiation-hardened electronics, because you cannot simply buy those chips off a shelf.

Nobody wraps a satellite in a lead vest. The mass would eat the whole launch. So machines get the other answer, and it is the strangely backwards one.

What a particle does to a chip

A transistor stores a bit as a small pool of electric charge. Charge present is a one. Charge drained is a zero. That is the whole trick.

An energetic particle from the Sun or from deep space is not a wave of heat. It is a single fast nucleus, and when it punches through silicon it knocks loose a trail of charge along its path. If that trail dumps more charge than the pool holds, the one becomes a zero. Nothing is broken. The number is just wrong now.

That is the mild version. The nastier one is a particle that switches on a parasitic path buried in the silicon and leaves it switched on. The circuit then shorts through itself, drawing current until something melts or somebody cuts the power.

This is a different failure from the one space is famous for. Heat in a vacuum is energy with nowhere to go. This is information going quietly wrong while everything still looks fine.

Every improvement makes it worse

Here is the awkward part. Everything that made your phone good made it worse at this.

Shrinking transistors is what made chips fast and cheap, and a smaller transistor holds a smaller pool of charge. A hit that a chunky old transistor would have shrugged off now flips the bit. Lower voltages, which is how modern chips stay cool, shrink the margin further. Faster clocks catch glitches that would once have faded harmlessly between two ticks.

So the consumer industry has spent forty years walking briskly in the wrong direction, for perfectly good reasons that have nothing to do with space.

Going backwards on purpose

The countermeasures are all the opposite of progress as normally understood.

Make the transistors bigger, so one particle cannot empty the pool. Run the clock slower, so a brief glitch dies before anything reads it. Build the transistors on an insulating layer so there is no buried path to latch. Run three copies of the same logic and let them vote, so one wrong answer loses. Have the memory continually re-read and repair itself. Add a watchdog that notices a hung processor and reboots it.

Then make all of that in tiny quantities, on production lines that are old because old is proven, and test it for years. That is why a flight computer costs more than a house and performs like a laptop from twenty years ago. It is not caution. It is the correct answer to the question actually being asked.

The rule flips close to home

The honest other half is that this is not true everywhere.

In low orbit, Earth’s magnetic field is still doing most of the work. A satellite a few hundred kilometres up is inside the shield, catching a fraction of what a probe to Jupiter catches. So the fleets circling overhead increasingly fly ordinary commercial parts, accept the occasional reboot, and buy their safety in a different currency: spare satellites instead of spare circuits.

That switch happened because launch got cheap. When replacing a satellite costs less than making it invulnerable, the engineering answer changes. “Old and slow” is not a law of space. It is what you do when the machine is somewhere nobody can reach it.

What we were measuring from

And then there is the part that should make anyone building this stuff uneasy.

For decades the field believed Earth’s response to a solar storm flattened out past a certain strength, as if the planet had a natural ceiling. Hardware was specified against that belief. This week’s re-reading says the ceiling was never physics. It was an artefact of measuring the wind from a spacecraft parked 1.5 million kilometres upstream, where the biggest readings are the ones most likely to be flukes.

The chips were designed correctly against a number that was wrong. So were the grid models, and the satellite fleets that carry your position fix, your weather forecast and the timing signal under a card payment. None of that failed. It was simply built to a threshold that came from standing a long way upwind and rounding off what arrived.

Most of what any of us build sits on a number somebody else measured, from further away than we picture, with more noise in it than the decimal places suggest.

03 · Lab · your turn

Spec the flight computer

Choose a destination and a chip, and feel why the safe machine in deep space is deliberately big, slow and old.

04 · Hope · carry this

The limit we trusted for decades turned out to be a measuring error, and the people who found it were the same ones who had been relying on it.

Across the beats