The last of a legendary rocket flew — to build a network on rockets that aren't ready

Space 4 min 15 sources

United Launch Alliance flew its final Atlas V 551, one of the most reliable rockets ever built, adding satellites to Amazon's internet constellation. The catch — the two newer rockets meant to finish the job are both grounded.

Key takeaways

The final flight of the powerful Atlas V 551 added satellites to Amazon's internet constellation, but the two newer rockets meant to finish the job are both grounded.
Amazon didn't build its own rocket — it bought launches from three companies, so the one thing it can't control, launch schedules, is the thing gating its whole plan.
Astronauts repaired the space station's main robotic arm on spacewalks, and astronomers found a rocky planet just 25 light-years away that might, or might not, hold an atmosphere.

At 12:30 a.m. Eastern on July 2, a United Launch Alliance Atlas V rocket lifted off from Cape Canaveral carrying 29 broadband satellites for Amazon’s Leo internet constellation [1][6]. It was the last flight of the Atlas V in its most powerful form — the “551” configuration: a five-metre nose cone, five strap-on boosters [1]. The first 551 ever flown launched NASA’s New Horizons probe toward Pluto back in 2006 [1]. This one, the 110th Atlas V overall, closed the book on that heavy version [1].

The Atlas V is one of the most dependable rockets in the business — a long record of getting where it’s pointed. Only six are left, all reserved for Boeing’s troubled Starliner crew capsule [1]. So a rocket with almost nothing left to prove is being retired, and the job of finishing Amazon’s constellation falls to newer, cheaper rockets.

Here’s the twist. Amazon didn’t build a rocket. It bought launches — 38 flights on ULA’s new Vulcan and 27 on Blue Origin’s New Glenn [1]. Both of those rockets are grounded right now, each working through its own failure investigation [1]. New Glenn is still sorting out its return-to-flight plan [4]. So the constellation Amazon is racing to switch on this year is leaning on the old reliable Atlas V and Europe’s Ariane 6 to keep adding satellites — while the future it paid for sits on the ground [1].

The numbers show why the pressure is on. Amazon Leo now has 396 satellites in orbit — 394 of them working, out of 398 launched since April 2025 [3][1]. Amazon wants to start commercial internet service by the end of the year, and it’s already signing customers: Hitachi will put Amazon Leo antennas on construction sites in Britain and Germany [1]. But a global network needs thousands of satellites, not hundreds. And Amazon can’t launch them itself.

Why bigger isn’t always the answer

A study out this week argues the industry’s instinct to build ever-bigger rockets may be wrong [5]. Bigger rockets carry more per flight, but they cost more to develop, fly less often, and hurt more when one fails or gets grounded — exactly the bind Amazon is in now, waiting on two big vehicles that can’t fly [5]. Sometimes a swarm of smaller, more frequent launches beats one giant stuck in the hangar [5].

Up on the station, astronauts operated on a robot

On the International Space Station, two astronauts spent the week doing something closer to surgery than repair. The station’s main robotic arm, Canadarm2 — the Canadian-built arm that grabs arriving cargo ships and moves gear around — had a fault [7]. Fixing it meant spacewalks: astronauts floating outside to replace parts of a robot that itself does the heavy lifting [8]. NASA said the crew “operated” on the arm across two outings, and by week’s end the repairs were done [8][9]. Astronaut Chris Williams was among those who carried out the work [10]. It’s a reminder that even in an age of robots, some jobs still need a human hand in a spacesuit.

A near-neighbour planet, and a close pass at an asteroid

Astronomers announced a rocky planet, GJ 3378b, orbiting a small red dwarf star just 25 light-years away [11]. That sounds far — light takes 25 years to cross it — but the Milky Way is 100,000 light-years wide, so in galactic terms it’s next door [11]. The planet sits in the “habitable zone,” the band around a star where liquid water could exist. Whether it actually holds an atmosphere is unknown — its star blasts it with radiation that could strip the air away [11]. A candidate, not a confirmation.

And on July 5 — today — a Japanese probe is set for one of the closest asteroid flybys ever attempted by a craft its size [12]. One scientist on the mission called it another “beast to put in the zoo of asteroids” [12]. Each close pass turns a distant dot into a mapped, measured world.

The telescope rescue is under way

Following on from last week: NASA’s mission to save its Swift space telescope — a 21-year-old observatory slowly falling toward Earth’s atmosphere — has launched [13][14]. A small spacecraft is on its way to boost Swift back to a safer orbit, an audacious attempt to rescue a telescope that was never designed to be caught [14]. Meanwhile in Chile, the new Rubin Observatory has begun what astronomers are calling the largest cosmic time-lapse in history — a decade-long survey of the entire southern sky [15].

02 · Lesson · why it matters

The one thing money can't buy is the clock

You can own the plan, the parts, and the customers, and still not own the moment it happens — and the thing you can't control is usually the thing everything waits on.

Amazon did almost everything right

Look at what Amazon has done to build its satellite internet network. It designed the satellites. It built the factory. It raised the money. It lined up customers before the service even exists — construction firms in Britain and Germany, ready to bolt on antennas.

By any measure of effort, this is a company that has bought and built its way to the edge of something huge. Nearly 400 satellites already circle the Earth.

And it is stuck. Not because it ran out of money, or ideas, or will. It’s stuck because it needs rockets to fly, and it doesn’t own a single one.

What you buy and what you rent

There’s a quiet line running through every big plan: the things you control, and the things you only rent.

Amazon controls the satellites. It rents the rides. It bought 38 launches on one rocket and 27 on another, from two different companies. On paper, that’s the plan finished. In reality, both of those rockets are grounded right now, each working through its own failure — and no amount of money on Amazon’s side makes them fly a day sooner.

That’s the trap of renting the critical step. You can pay for it. You can sign the contract. But you can’t reach into another company’s factory and fix the problem keeping their rocket on the ground. Your schedule now lives inside their schedule. Their bad week is your bad week.

The lever you don’t hold

Here is the part worth carrying past today. In almost any plan, there’s one lever you don’t hold — and it’s rarely the one you worried about.

Amazon worried about the hard, visible things: can we build the satellites, raise the money, win the customers. It solved all of those. The thing that has it waiting isn’t any of them. It’s a launch schedule set by other people, gated by a failure investigation nobody can rush.

That’s the pattern. We pour our attention into the parts we can grip — the ones we’re good at, the ones that respond to effort. And the thing that actually decides when we arrive is often the one part that doesn’t care how hard we push. It moves on its own clock.

Why the old reliable rocket still matters

Notice what’s holding the plan together in the meantime. Not the shiny new rockets Amazon paid billions for. It’s the old one — the Atlas V, a rocket being retired precisely because it’s expensive and dated, flying its last heavy mission to keep the satellites going up while the future stays grounded.

The thing you were about to throw away is doing the work the thing you invested in can’t yet do. That’s not an accident. The proven, unglamorous option earns its keep exactly when the ambitious one stalls. Reliability is boring right up until it’s the only thing still moving.

We are all waiting on someone else’s clock

This isn’t only a story about rockets. Look at your own plans. The move that waits on a mortgage approval. The launch that waits on one supplier. The trip that waits on a visa office. The recovery that waits on a body healing at its own pace.

You can be diligent about everything you touch and still be held by the one step that answers to no one — a court, a queue, a season, another company’s engineers. The most careful plan in the world still runs partly on clocks other people wind.

Amazon can see the whole board. It has the satellites, the money, the customers, the contracts. And it is standing at the pad, watching, waiting for a rocket it cannot rush. Seeing that clearly doesn’t make the waiting shorter. It just makes it honest — a reminder of how much of any plan, our own included, was never fully in our hands to begin with.

03 · Lab · your turn

Own The Clock

Choose whether to build or rent the critical step, then feel how much of your plan you controlled when a year goes wrong.

04 · Hope · carry this

The rockets are grounded only because someone caught the problem before it flew, and the satellites keep going up because a hundred separate teams still know how to hand the work off to each other. Progress rarely runs on our schedule, but it does keep running.