- Launch roughly on time, after a scrub yesterday. (Sounds like the scrub was due to ground equipment, most notably the water system.)
- Initial ascent was good, but then one engine on the booster went out.
- Relight of the booster's engines after stage separation for the boost back burn failed. Engines did light again for a landing burn, but seems to have hit the water harder than expected and was very off target.
- Starship lost one engine shortly after stage sep. Turned into an unintentional test of engine out capability. It made it to space.
- Some weird motion and lots of off-gassing after engine cut-off, with uncertainty about if it actually got a good orbital(ish) insertion. Seems to have been benign, with the motion being a weird slow flip to the orientation for payload deployment.
- Test deployment of dummy payloads was successful, including a couple with cameras to look back at Starship.
- An in space engine relight test was skipped, presumably due to the issues during launch.
- Re-entry to over the Indian Ocean seemed to go really well. Nothing obviously burning or falling off. The amazing views of the plasma during re-entry, something never seen live before starship, are now routine.
- Starship did a maneuver to simulate how they'll have to go out over the gulf and back to the landing site.
- Nailed the target, evidenced by views from drones and buoys. Soft landing before falling over and giving us a big (expected) boom.
As far as overall progress from previous test flights goes, they're at least treading water while making many large changes. I think they were hoping to try for a tower catch and actually going orbital for next flight, but I highly doubt that now. The boostback burn failing was the largest failure, with the engine failure on Starship being a close second. Good performance despite engine out seems to be an unintentional success.
Good summary. The booster appeared to hit the water at 1400 km/h (a bit under 900 mph) so not really survivable :-). Engine out on ship seems to left them with just enough fuel to land but not enough to do the hover thing (simulates being caught by chopsticks). They notched it down to two engines (vs planned 3) on the landing it seems?
Basically if they can figure out the engine issues, it looks like they should be able to do a full end to end flight. That's reasonable progress. Given the IPO this was a pretty important flight and I don't think they hurt it (like blowing up on the launch pad would have). So their one step closer it seems.
I doubt it since many of the booster engines didn't seem to relight, the location of touchdown wasn't near any pre-positioned cameras (if there were any).
The videos are great!, but the rest of it is never going to work lol, just never. Even without a rethink about how to get heavy payloads to another planet this is still good entertainment.
Even if it landed perfectly how is it going to be rapidly reusable with all those tiles breaking and needing repair? Then if that problem was magically engineered-away through some sort of materials science breakthrough, it still makes more sense to me to keep your big ships in a space staging area and your smaller ones as atmospheric gophers.
Could you tell me more? I suppose a heavy two-stage rocket is not optimized from the point of view of the rocket equation, but I know nothing about this field.
Shuttle's tiles not being durable as hoped is what killed it's turnaround time.
The problem was never solved and turned what was supposed to be a few days into weeks or months. Every mission the shuttle had to go back into the assembly building and have all tiles inspected and potentially replaced.
Its turn around time is ridiculous, it has to be maintained with specialized equipment/hangers, along with external contractor assistance.
Compared to the Gripen, as an example, which can land on a freeway and be up in the air again in a few minutes.
One was designed to be used in war, in desperate scenarios, with no ability to coddle it. The other, the F-35? Is designed around milking the taxpayer as much as possible, and employing people in as many politician's states as possible.
The shuttle was like that, I think. Which is really sad.
Agreed and specifically in the case of the Gripen the “test condition” was “Needs to be serviceable by a few conscripts working under the direction of one person who knows what they are doing”.
It’s an extremely different design goal, the US doesn’t mind exotic weapons that require exquisite (and expensive) methods of servicing, they have the budget and the assumption that a well equipped air field will be immaculately maintained.
Meanwhile the Mig-29 designers assumed it’d operate from damaged/poorly maintained fields, so on the ground you can shut the primary air intakes and it uses ones on top of the plane to get air, drastically reducing the FOD risk on taxi/takeoff.
I do wonder how well the F-35 would fare in an actual shooting war against near peers when all the peacetime assumptions breakdown.
I think there's also some exaggerations about the differing highway landing capabilities of various aircraft. [1] is a video showing Eurofighter, F/A-18 and F-35 all landing on a highway in an exercise. Capability with stores and fuel load is another thing but I've read material that doesn't find the contemporary aircraft drastically different in that regard. Now, maintenance hours per flight hour and general operability certainly are interesting topics and there could be large differences.
I mean ... step 1 is probably fixing the part where it lands in the ocean, falls over and explodes. Once they've done that and can get their hands on the tiles I'm guessing they can continue to iterate there until they get a more easily reusable design.
“seems to have hit the water harder than expected and was very off target.”
SpaceX’s people were saying it was on target, and it seems to have landed in about the same position relative to the camera buoy as previous flights. I don’t think there’s any evidence to call it off target. The landing and toppling looked the same as previous flights too.
If we look at the venting from the propellant tank (around T+16:15) it looks thick white closer to the vent, becoming more transparent and blue as it expands. That's just sunlight scattering on the particles and density fluctuations in the flow.
A good cold gas thruster produces a lower density, more expanded flow, which looks blue for the same the reason the sky looks blue.
One can compare this to the exhaust from various Falcon-9 engines and thrusters when it is illuminated by the sun on the backdrop of the night sky:
https://youtu.be/JRzZl_nq6fk?t=193
Did the landing burn light two engines as expected? It happened fast, but the graphic made it look like only one lit. If that’s true, that would be impressive as only lighting two was meant to be a test. At least according to the live stream hosts.
Booster is a totaly new rocket.
It did launch with acarity and by any other standard did well, but the failure to relight could be anything, but I am going for the giant fuel feeder tube bieng the failed part, based on nothing more than how tickled they were with it,and tank baffling bieng a dark art.
The slosh of the fuel durring the flip is going to produce an internal tidal wave, lots of stuff gets "tested" there.
Lots of engine failures. Doesn't exactly bode well for a company looking to go public immediately. One of the engine failures was not on the booster but Starship as you noted, and that is a bit unexpected. I don't think they have spoken about it being equal in capability with one engine out, right? Those engines don't move around to compensate IIRC.
Not sure how you come to that conclusion. The capabilities can overcome loss of engines. The fact it was successful with loss of engines shows it is working as designed.
No, it just means the mission happened to be salvageable because of its parameters. The booster is designed to have engines out and can compensate because it has so many engines and many of them are on gimbals. On starship, the vacuum engines aren’t on a gimbal. I’m not sure how it could compensate for one of three engines being out.
Some are on a gimbal and they specifically talked how others gimbaled out a bit to compensate. This is specifically something they designed in and not an accidental lucky save. In this flight they didn’t intend to test “one engine out” feature but it worked out that way.
See my other comment. The vacuum engines are NOT on a gimbal. None of them. The sea level engines on starship and several of the engines on the booster are on a gimbal. But not the vacuum engines for space.
EDIT: I cannot reply further in this thread, but my understanding is that the non vacuum engines are not intended to stay lit throughout the orbital flight in a typical mission. If they are, they can gimbal and compensate.
> The vacuum engines are NOT on a gimbal. None of them
I said some raptor engines are on a gimbal, not vacuum engines.
To be precise, the three central engines can gimbal up to 15 degrees. That can control the thrust vectoring when an engine fails, and that’s what happens in the last flight.
Since the flight already happened and we know it didn’t spin out of control (unless you imply their diagnostic and telemetry was completely off and the engine was actually on) something must have compensated for the failure. It wasn’t magic, it was in fact the central 3 engines that did that.
You may be confused because those are called sea level engines, but that doesn’t mean they can’t work in vacuum.
I think previous comment means "on a gimbal" as in "angled at a non completely prograde direction" (presumably angled such that each engine points through center of mass so that none of the engines impart a torque)
The person you’re replying to is trying to play rhetorical word games.
The upper stage has six engines. The outer three engines are “vacuum engines” (optimized for operation in space). The inner three engines are “non vacuum engines” (optimized for operation in the atmosphere, at sea level).
The outer three vacuum engines are not gimbaled, but the inner three sea level engines are. Thus, it is completely accurate to say that they gimbaled some of the engines to compensate for the engine failure.
That’s for the booster (the big lower part) not for starship (the upper part that continues to space). They were surprised to have a vacuum engine out. In space there’s no atmosphere so you can’t use fins or wings to change direction. And if the engines can’t move around, you only have thrust and gravity and the tiny attitude adjusters to direct your ship.
You're simply wrong. The non-vacuum-optimized engines on the upper stage are still functioning in a vacuum, and their ability to gimbal to offset the loss of one of the vacuum-optimized engines was planned for.
Watch the stream again starting with the ship burn. They explicitly said they have engine out capability on ship. The sea level engines on the ship are running and gimballing.
The takeoff looked almost normal but I noticed a slight drift from vertical, likely because one of the engines was dead or dying. Overall the V3 is supposed to be an upgrade but actual progress is more or less stalling compared V2.
It is supposed to tilt away from the launch tower immediately, you can see this on previous flights. This keeps the engine plume away from the chopsticks and top of the launch tower.
The best part of this flight was seeing the full reentry with no visible hot spots or burn through like we've seen on every previous reentry of Starship. Seems like they have the heat shields really nailed.
Yes, reuse of the heat shield has been the biggest question mark of the whole program and this is by far the best result of any launch so far. This is the first time it looks plausible that you could consider reusing the heat shield.
I am just delighted that SpaceX continues with the "good enough" pace of development here, at least at these phases. Rapid iteration of build, test, learn, and improve rather than wait for perfection.
They are willing to have "negative outcome learning experiences" to gather data quickly. and, of course, data, data, data.
I like it because I know what insane amount of red tape has built up to do anything similar in a Gov (any Gov).
Shame they're risking that ability with the IPO. We've seen how irrational and ignorant stock traders are from other publicly traded space companies. Even scrubbed launches cause the price to dip.
> We've seen how irrational and ignorant stock traders are from other publicly traded space companies.
Absolutely true, but ignorant stock traders making irrational trades only matters if company management pays attention to them. Musk will maintain complete control of SpaceX even after the IPO, so he can focus on long-term value rather than short-term ups and downs.
Of course, over time, if more shares are issued, this may change.
The views from Ship's engine bay looked rather ominous -- with the red glow visible in multiple places, and something venting furiously from the broken engine. It was a pleasant surprise that the ship did not explode and not only that, but it even landed exactly on target. Guidance system software engineers have done a very good job!
The booster not completing the return part of the flight was disappointing. They had a similar incident in one of the previous flights, when they tried to maneuver the booster too aggressively immediately after stage separation which caused problems with the fuel supply. If it was something similar this time, it might be solvable by changing just a few details of the maneuver. So, maybe it is not that huge of a deal.
There were many cool things in the webcast, from them showing the catamarans that are deployed at the landing site, to the views form the cameras on-board of the "satellites". The first few minutes after liftoff were just amazing visually.
My favorite part of this launch that others haven't already mentioned: during reentry, the dummy payload satellites were visible burning up behind the ship!
This was good forward progress (V3 mostly worked, clear improvements on heat-shield, near-final Starlink deployment system).
Is this enough progress to keep a 2028 crewed landing? Don't know.
I'm curious whether they are going to try to recover a Starship before trying for in-space refueling (or the reverse). Either way, I think both have to work before they can try for an uncrewed lunar landing (presumably in 2027).
The big question is re-usability. How close are they to relaunching a Starship? They may not know for sure until they can get one back intact. If they can launch at least once a month, maybe they'll make it.
If they can re-fly a Starship this year AND demonstrate in-space refueling, then 2027 can be all about an uncrewed landing attempt. That would make me feel good about a 2028 crewed landing on the moon.
> If they can re-fly a Starship this year AND demonstrate in-space refueling
I'd bet that they'll not try in-space refueling before they demonstrated in-space relight of an engine. So they need to fly at least twice. Or even thrice because to demonstrate refueling you need two Starships in orbit.
The amount of data they must have at this point running so many of those raptor engines has got to be insane...
at least 300+ engine launches now -- wow.
Five years ago SpaceX reported that they had 30000 seconds of test firing time on the Raptor, over 567 engine starts. Since them the program accelerated dramatically. Well over one thousand engines had been produced, and on an average day at McGregor test facility the Raptors are fired for about 600 seconds. That would give about a million seconds over five years. That's a lot for any engine development program.
A typical test stand would have maybe a thousand channels of relatively slow data (pressures, temperatures, flow rates, valve states, etc), and maybe up to a few hundred of channels for essentially audio data from vibration sensors. This amounts to sub-gigabit per second data rate overall.
If very high speed video / multiple video cameras are used, this could generate massive data rates, but unless something interesting happens it is not clear how important this data is.
In flight, the telemetry data rate from the entire Falcon-9 used to be measured in megabits per second per stage, plus the video stream. It was not a huge amount of data. Presumably now with Starlink they send a lot more telemetry from Starship, but in flight the engines typically have far, far fewer sensors compared to the ground testing.
> Sprint accelerated at 100 g, reaching a speed of Mach 10 (12,000 km/h; 7,600 mph) in 5 seconds. Such a high velocity at relatively low altitudes created skin temperatures up to 6,200 °F (3,400 °C), requiring an ablative shield to dissipate the heat. The high temperature caused a plasma to form around the missile, requiring extremely powerful radio signals to reach it for guidance. The missile glowed bright white as it flew.
It seems to give the booster a real kick - what's that do to turbo's and fuel movement?
You've got hot exhaust onto cold cryo fuel tank header?
You've got to carry more mass in terms of protection for the tank?
Is doing MECO and then push and then get 100 yards apart or something before second stage / ship engines kick on a big enough penalty to justify all the extra complexity?
I wonder if the hot separation was supposed to be that hot. Going at mach 5 and doing a quick U turn while there was some weird orange color on the side of the Super Heavy, then (possibly?) losing most engines from it seemed extra chaotic
They had a similar issue with the v2 booster the first time, flipped it so hard it damaged the downcomer and they had to change it up a bit. Seems likely here, that thing thats as big as a building flipped ends pretty quick and is partially filled with liquid.
Oh man, so glad I stayed up to watch it. Kind of a rough start (but it's the 1st flight w/ new redesign, new engines, etc), had an engine out on both booster and ship, but the views were absolutely worth it. They managed to get the last satellite to connect to starlink and download the footage of the ship in orbit. Even with an engine out, the ship managed to reach orbit, deploy all the satellites, re-enter, flip and soft splash into the ocean, near a buoy! And on top of that we got the drone views of the landing. Fucking spectacular views.
I'm guessing / hoping that the engine outs we're planned, or that they ran the engines with slightly different parameters to test them. If it's just unreliability then it might be a hard problem to solve.
> If it's just unreliability then it might be a hard problem to solve.
It might, but it certainly helps having a ton of them around. Given that they used 42 of them today and 2 failed in some fashion, we'll call that a 1:21 failure rate. On a more typical rocket with say 10 engines (eg falcon 9), there's a good chance they wouldn't have seen the same failure till flight 3.
It’s something like up to 6 can fail and it keeps going, seems pretty good. I know they did some stuff like remove a heat tile to get failure feedback, wonder if engine was planned or accidental
Which is true, but at the same time: this is Starship Flight 12.
The whole point of Starship is that it's a reusable vehicle with easy turnaround and quick maintenance. And in particular it's supposed to be different than the other reusable vehicle with easy turnaround and quick maintenance, which turned out to be sort of a boondoggle.
Yet, they've now hand-built and destroyed twelve of these things across multiple redesigns, and it still hasn't completed its design mission once. In fact basically every launch has unexpected major failures.
As poor as its safety record ultimately ended up being, the shuttle launched successfully on its very first try. And we only had to hand-build five of them. And lost two, sure, which is still a lot less than twelve.
Yes yes, I understand that iterative design has merits and that the ability to rapidly prototype and try things in the stratosphere allows for less conservative tolerances and better ultimate performance.
But does it really take 13+ tries?! At what point to we start wondering if we have another boondoggle on our hands?
The part that makes no sense to me is why they are going starship scale rather than falcon 9 scale. Had they done their prototyping on a rocket with 9 engines on the first stage and 1 on the second, they could have gotten to raptor 3 (and a falcon 9 replacement) while blowing up way fewer engines, launch complexes, etc. There's a reason Spacex started with the falcon 1 rather than the falcon 9. It's a lot cheaper to blow up fewer engines and smaller rockets while you're developing a new rocket engine.
The point was more that there is a point where (to borrow the software terminology) "iterative design" becomes "death march". Trying a few times in the early days and being willing to throw stuff out and start over is a powerful tool.
I think blowing up a handful of rockets is a fine idea. But at some point you have to ask yourself if it will ever work? Why are we on a another engine redesign? Why is this the third iteration of the second stage? How many more?
And what number is that point? Six? Nine? I'm thinking thirteen may be getting into the danger zone.
In a somewhat similar situation Sergey Korolyov stopped his colleague in front of the Party officials asking a similar question and explained: "We are exploring terra incognita, this is the process of getting knowledge". He was sort of right - even though there were many specific engineering problems, and many of those were rather solvable, especially in hindsight, overall process was stepping into the unknown.
Here we have a cutting edge rocket design - scale, sophistication of engines, design goals - and a commercial evaluation, which path would get to the intended success cheaper. NASA doesn't like public embarrassments, and, as Henry Spencer reminds us, when failure is not an option, the success could be quite costly. So NASA spends billions and many years for a fragile system. If the goal is an airline-like operations, the design should be thoroughly shaken up. It's known that no simulation, no static testing can equate the actual flights in the ability to get the data best describing what conditions the system will encounter in real use. And also, given the industrial scale of Starship production, each flight hardware costs way less than if we'd built them manually, in quantities justifying naming each unit separately.
In Soviet Union, where rocket departments were part of artillery, the testing with actual launches seemed logical. In this case the approach to run a massive test flight program seems logical too, and we can't complain about the lack of progress - first Starship had way less capabilities and performed way worse. In USA we had more than 1000 tests for injector head for F-1 engine in Apollo program, and this number was justified at that time. Starship is way bigger - but the progress is also undeniable, and it would be odd to stop test flights now, when the 3rd iteration of design looks promising.
So, while we can't pin a particular number of tests, I don't think we should worry yet. This year and the next one should be important for Starship program, given SpaceX commitments to help NASA Artemis. If we won't have orbital Starship then - we can come back to this question.
So what if they blow up literally 100 rockets, if they can eventually perfect it faster and more cheaply than the traditional approach, recently typified by SLS.
SpaceX have already proven that the iterative approach works with Falcon 9, literally the most successful rocket program ever. SpaceX have also proven that this specific Super Heavy/Starship rocket design isn’t a dead end. Criticising them for failing to succeed in the future is a valid but uninteresting opinion.
It’s cheaper and faster to make in volume. It doesn’t require nearly as much shielding, because it’s less fragile, which saves a lot of weight. The engine itself is lighter. And on top of that, it develops more thrust, at higher fuel efficiency.
The net result is cheaper and lifts significantly more mass to space, which significantly drops the cost per kg to orbit.
It already worked, they’re making it much better, and getting it ready for a level of mass production that we’ve never seen anything close to in the space industry, even from SpaceX. They are much more ambitious than I think people who haven’t been watching them closely understand. The US grid is 1.4 TW of generation, they’re aiming to put up 1 TW of AI compute every year. Maybe they’ll stop well short of that, but their stated goal is insanely ambitious.
v3 is the first version that was made with the intention of being used for actual payload delivery. The versions before were about testing and proof of concept.
But all of those 12 launches happened in just 3 years, and cost a tiny fraction of other major spaceflight development programs.
For reference, SLS has been in development for 5 times as long, and cost 15-20 times as much, as Starship, and they still haven’t landed people on the Moon, which has been one of the stated goals since the Constellation program in 2005.
I don’t see how the number of failures matters if the end result still happens faster and cheaper than anything else.
Recent SpaceX IPO filings put that 'tiny fraction' at about 1/3 or 1/2 of SLS. $15B total investment with about $4-5B of that figure from US gov. Is starship more than 1/2 or 1/3 of the way to a human rated Artemis II style mission? The main reason starship costs less to test (apart from the SLS jobs program baggage) is because of design choices which prevent it from performing such a mission without significant further tech development.
'5 times as long' is dubious too. SpaceX claims to have been working on the design since 2012 vs 2011 for SLS. Ultimately though the start date of a complex program is not well defined, as early conceptual design stages can take years without leaving the drawing board. Government needs to put a start date on such efforts for legal/budget reasons, but a private company does not.
Also relevant - SpaceX has been given a lot of tech and expertise from NASA at a tiny fraction of the cost and time it would have required them to develop it themselves. Therefore, the costs of NASA programs like space shuttle actually includes some of the development costs of SpaceX.
Both programs pale in comparison to Saturn V, which was faster, cheaper, and more technically demanding at the time.
Moreover the two lost shuttles included human lives. Better to blow stuff up with demo payloads now before sending up large contracted payloads or worse human beings!
I couldn’t believe my ears when I first heard that the second ever flight of SLS was going to be crewed.
It worked out in the end, but I can’t imagine being so confident in a new system, no matter how much money and brainpower has been spent to make it safe.
That's undeniably true. Nonetheless "Better than the shuttle, which sucked" isn't the design goal.
The question is not even just "is it better to blow up 12 Starships?", which would probably still be true. It's "Why isn't Starship working yet?" and the implied "Maybe Starship sucks too?!".
Now the flaps don't melt! The tiles don't fall off!
It's a major overhaul of the design they've been working on for a long time. There was talk of v3 fixing the problems in early v2 test flights. The booster is v3 as well which presumably is why they had some problems. I believe this is also the first time they flew the v3 engines with the plumbing fully integrated in a single piece housing they 3D printed.
Big takeaway for me is that the reentry and “landing” of Ship looked great. For the first time, it felt like they’re really on the path to achieving upper stage reuse. That was always the biggest “reach” of the entire program in my view, and today they took a major step forward.
Is it disappointing that they had a couple of engine outs, and also trouble with the booster relight? Sure. Do I have even a little doubt by now that they can fix these problems? None whatsoever.
Having a faultless payload deploy and a pinpoint landing after losing a whole vacuum engine (one of 3) so early was an unexpectedly amazing performance. I suppose they gimballed the inner non-vac engines to the max and burned longer, next level adaptability.
Most obvious improvement was having no re-entry heating problems, secondmost was deploying with zero issues and with a faster pace. It appears they decided to pause the "horizontal" movement of the pez dispenser before a final push away, probably to avoid vibration causing those "bonks" on the payload door, like we had once before.
It's worth remembering that, according to SpaceX's own filings, they've spent >$15 billion on the Starship program thus far with more to come. And SpaceX is burning cash still, particularly because Elon Musk bailed out his own bad decisions with Twitter and xAI with SpaceX stock, basically.
Flight 12 was a relative success. Some engines failed to light but that's an unintended good test. Rockets are typically designed such that they can have a certain number of engines fail and still achieve their mission.
At this point, the entire SpaceX project is a bet on telecommunications services, specifically direct-to-satellite handheld Internet. That's the only market that will recoup the program costs.
We don't have exact figures for the current true cost of a Falcon 9 launch factoring in reuse but many think it's somewhere betweenm $10 and $20 million. Well, SpaceX has spent 100 F9 launches on Starship so far and that's how you have to look at it. Say F9 is $20M and Starship once it starts launching Starlink is $10M that's 150-300+ launches just to break even.
You might be tempted to say there are other missions for Starship but there really aren't. Satellites aren't that bug, as evidences by there being ~1 Falcon Heavy launch per year (usually for the military and/or to geostationary orbit AFAICT). You can't economically put multiple payloads in one Starship because they all have different orbital parameters.
F9 is rated for human spaceflight. It's a long road for Starship to be certified for human spaceflight. SpaceX hasn't even begun to test in-orbit refuelling yet. Gases are weird in microgravity.
F9 is the cash cow funding all this and that too might go away if Blue Origin or one of the other wannabes ever gets a reusable launch platform to commercial operation.
There are big launches like interplanetary missions but those are few and far between.
It would be fascinating if what ends up dooming SpaceX is actually Twitter.
> At this point, the entire SpaceX project is a bet on telecommunications services, specifically direct-to-satellite handheld Internet. That's the only market that will recoup the program costs.
There's also a military angle here. I'll leave it as an exercise for the reader to look into Musk's history with Michael D. Griffin from the Reagan SDI/'Star Wars' program.
It’s an awkward comparison, but F9 can deliver a payload to orbit at a slightly lower price per kg than a Tomahawk missile can deliver it to a target. Starship would be MUCH cheaper if the economics works out the way that SpaceX would like it to.
Obviously a few hundred kg of payload in orbit are not equivalent to the same payload delivered directly to a target.
You don’t need very many kg delivered to target if it’s plutonium. The SDI program had the idea was that if you parked enough defensive weaponry in orbit then maybe mutually assured destruction wasn’t something you had to worry about. The only problem was that getting all that mass into orbit was prohibitively expensive.
Then the deputy director of the program met a young man named Elon Musk, and the rest is history.
> At this point, the entire SpaceX project is a bet on telecommunications services, specifically direct-to-satellite handheld Internet. That's the only market that will recoup the program costs.
I seriously doubt that. Just for example, mining a single asteroid has the potential to flood the market for any number of metals. I don't pretend to know how expensive it would be to achieve that in practice; my point is that there are quite a few different ways to recoup program costs at some handwavey point in the future.
If there were infinite gold bars just sitting on the surface of the moon, it wouldn't be economical to go collect them and bring them back to Earth. No matter how expensive you think any metals are here on Earth, the cost of launching vehicles, rendezvousing with said metals and bringing them back to Earth makes it uneconomical.
An asteroid is much, much further than that but more important than distance is the delta-V required for change its orbit to reach an Earth orbit. So you not only need to get there, which, as discussed, requires in-orbit refuelling with Starship (or any vehicle), but you have to carry all the fuel you need for the orbital burn to bring it back. The rocket equation just kills this immediately.
You really hope you have to get incredibly lucky that an metallic asteroid is on a near-intercept course with Earth that is just shy or going into orbit. The odds for that are, well, astronomical.
xAI is burning through $1 billion a month [1]. With Anthropic as a customer, it's basically an argument that we're losing money on every transaction but we'll make it up in volume.
> That's the only market that will recoup the program costs.
No. If it is just $15B I can think of dozens different usecases ranging from military applications(fast transportation, it is the cheapest ICBM) to asteroid deflection to moon mining to science applications to space datacenter.
Are you seriously thinking $15B is big? Artemis by comparison has spent $93B and has cost of $4B per launch.
Another flight with many explosions and a trivial payload. Trial and error, trial and error. At least these million monkeys have upgraded from typewriters to something more fun.
Had to check this one and you're indeed right: one death in 2014[0], and one death earlier about a week ago (May 2026)[1] (as well as 600+ workplace injuries [0]).
This incremental progress, far smaller improvements than planned, has put them so far behind schedule I'm not confident this design is any good. Still haven't done orbit. This launch was not a smooth launch. SLS by contrast seems to work. Why did nasa contract SpaceX for the lander. The whole plan is bad.
Are you serious? They just launched a completely revamped version of Starship from an entirely new pad, and still hit almost all of their planned milestones while demonstrating that the design is reliable enough to handle a missing engine.
Two successful test missions over 15 years using a dead end rocket and ship design for $50+ billion is not something to be proud of either.
Starship's design is so far beyond where the rest of the world is that even if it takes another 3 years of iteration to perfect it will still be by far the best rocket in the world for many years afterward, to the point where it would hardly make sense to choose to launch on any other existing or currently in development rocket for any reason other than political ones.
There's plenty of finicky systems which go on to be good systems with a lot of work. Some things are just hard, a lot of the time you just don't see them being hard so publicly.
That's software engineer "I'll simplify it!" logic and the score of it's successes is far too low for how common it is.
If a design with a bunch of modifications works, then it's a good design. Thinking you need to clean sheet redesign everything is how you get Second System Syndrome.
Summary from my watch:
- Launch roughly on time, after a scrub yesterday. (Sounds like the scrub was due to ground equipment, most notably the water system.)
- Initial ascent was good, but then one engine on the booster went out.
- Relight of the booster's engines after stage separation for the boost back burn failed. Engines did light again for a landing burn, but seems to have hit the water harder than expected and was very off target.
- Starship lost one engine shortly after stage sep. Turned into an unintentional test of engine out capability. It made it to space.
- Some weird motion and lots of off-gassing after engine cut-off, with uncertainty about if it actually got a good orbital(ish) insertion. Seems to have been benign, with the motion being a weird slow flip to the orientation for payload deployment.
- Test deployment of dummy payloads was successful, including a couple with cameras to look back at Starship.
- An in space engine relight test was skipped, presumably due to the issues during launch.
- Re-entry to over the Indian Ocean seemed to go really well. Nothing obviously burning or falling off. The amazing views of the plasma during re-entry, something never seen live before starship, are now routine.
- Starship did a maneuver to simulate how they'll have to go out over the gulf and back to the landing site.
- Nailed the target, evidenced by views from drones and buoys. Soft landing before falling over and giving us a big (expected) boom.
As far as overall progress from previous test flights goes, they're at least treading water while making many large changes. I think they were hoping to try for a tower catch and actually going orbital for next flight, but I highly doubt that now. The boostback burn failing was the largest failure, with the engine failure on Starship being a close second. Good performance despite engine out seems to be an unintentional success.
Nice work by SpaceX engineering.
Good summary. The booster appeared to hit the water at 1400 km/h (a bit under 900 mph) so not really survivable :-). Engine out on ship seems to left them with just enough fuel to land but not enough to do the hover thing (simulates being caught by chopsticks). They notched it down to two engines (vs planned 3) on the landing it seems?
Basically if they can figure out the engine issues, it looks like they should be able to do a full end to end flight. That's reasonable progress. Given the IPO this was a pretty important flight and I don't think they hurt it (like blowing up on the launch pad would have). So their one step closer it seems.
Is there videos of booster crash?
I doubt it since many of the booster engines didn't seem to relight, the location of touchdown wasn't near any pre-positioned cameras (if there were any).
The videos were incredible. My favorite part was watching the booster flip in such clarity. Normally we don't get full view of it, let alone 4k.
The videos are great!, but the rest of it is never going to work lol, just never. Even without a rethink about how to get heavy payloads to another planet this is still good entertainment.
why won't it ever work?
Even if it landed perfectly how is it going to be rapidly reusable with all those tiles breaking and needing repair? Then if that problem was magically engineered-away through some sort of materials science breakthrough, it still makes more sense to me to keep your big ships in a space staging area and your smaller ones as atmospheric gophers.
Small ships are less efficient, especially leaving the gravity well. Thats the whole point
Could you tell me more? I suppose a heavy two-stage rocket is not optimized from the point of view of the rocket equation, but I know nothing about this field.
While we're downing this I'd like to add I think the "I've written X in Rust" posts are also cringe :D
Good summary. I was pleasantly surprised that they nailed the re-entry target even after the ascent engine problems.
The re-entry itself looks amazingly smooth compared to V2. TBD whether it's good enough for re-usability (much less rapid re-usability).
But Flight 12 was definitely forward progress.
I'm concerned about the cracking clearly visible on the heat shield tiles. It doesn't bode well for rapid reusability.
I thought the tiles were designed for easy replacement, so not a big concern with replacing cracked ones.
The tiles ablate. The shuttle returned from every mission with missing tiles.
Shuttle's tiles not being durable as hoped is what killed it's turnaround time.
The problem was never solved and turned what was supposed to be a few days into weeks or months. Every mission the shuttle had to go back into the assembly building and have all tiles inspected and potentially replaced.
Shuttle tiles were also unique per position and starship tiles have a few base forms that are interchangeable
I would also believe that a robot could inspect and replace tiles a lot faster than humans.
Total 6 shuttles built over 35 years. SpaceX already crashed 12 over 5 or so years.
Obviously doesn’t guarantee they’ll find solution, but fast iteration will definitely help.
Well, every mission that it returned from it had missing tiles. That is not the same thing as returning from every mission.
The shuttle required long expensive refurbishment after each flight.
Just made me realise, this is just like the F-35.
Its turn around time is ridiculous, it has to be maintained with specialized equipment/hangers, along with external contractor assistance.
Compared to the Gripen, as an example, which can land on a freeway and be up in the air again in a few minutes.
One was designed to be used in war, in desperate scenarios, with no ability to coddle it. The other, the F-35? Is designed around milking the taxpayer as much as possible, and employing people in as many politician's states as possible.
The shuttle was like that, I think. Which is really sad.
Agreed and specifically in the case of the Gripen the “test condition” was “Needs to be serviceable by a few conscripts working under the direction of one person who knows what they are doing”.
It’s an extremely different design goal, the US doesn’t mind exotic weapons that require exquisite (and expensive) methods of servicing, they have the budget and the assumption that a well equipped air field will be immaculately maintained.
Meanwhile the Mig-29 designers assumed it’d operate from damaged/poorly maintained fields, so on the ground you can shut the primary air intakes and it uses ones on top of the plane to get air, drastically reducing the FOD risk on taxi/takeoff.
I do wonder how well the F-35 would fare in an actual shooting war against near peers when all the peacetime assumptions breakdown.
I think there's also some exaggerations about the differing highway landing capabilities of various aircraft. [1] is a video showing Eurofighter, F/A-18 and F-35 all landing on a highway in an exercise. Capability with stores and fuel load is another thing but I've read material that doesn't find the contemporary aircraft drastically different in that regard. Now, maintenance hours per flight hour and general operability certainly are interesting topics and there could be large differences.
1: https://www.youtube.com/watch?v=kKbgtixpfIc
I mean ... step 1 is probably fixing the part where it lands in the ocean, falls over and explodes. Once they've done that and can get their hands on the tiles I'm guessing they can continue to iterate there until they get a more easily reusable design.
That part was intentional
Dang, a random HN user solving all the world's problems yet again, what would humanity do without you random HN guy?
“seems to have hit the water harder than expected and was very off target.”
SpaceX’s people were saying it was on target, and it seems to have landed in about the same position relative to the camera buoy as previous flights. I don’t think there’s any evidence to call it off target. The landing and toppling looked the same as previous flights too.
You've mixed up the ship and booster.
SpaceX does an excellent job at videography. Sad that Nasa flew its Artemis mission with potato cameras.
Hey, we have everyone watching, our funding might in part depend on interest and awe…
Not just space-potatoes… but missed the separation shot on the live feed. How in the hell!?
I think the ship really punted the booster during stage separation. And caused the boost back failure from sloshing.
Also I think Ship now has methane thrusters on it. They were operating with a clean blue flame in short purposeful bursts.
If we look at the venting from the propellant tank (around T+16:15) it looks thick white closer to the vent, becoming more transparent and blue as it expands. That's just sunlight scattering on the particles and density fluctuations in the flow.
A good cold gas thruster produces a lower density, more expanded flow, which looks blue for the same the reason the sky looks blue.
One can compare this to the exhaust from various Falcon-9 engines and thrusters when it is illuminated by the sun on the backdrop of the night sky: https://youtu.be/JRzZl_nq6fk?t=193
The final issue that led to the scrub was that a pin that held back the QD arm got bound and would not release.
Did the landing burn light two engines as expected? It happened fast, but the graphic made it look like only one lit. If that’s true, that would be impressive as only lighting two was meant to be a test. At least according to the live stream hosts.
Booster is a totaly new rocket. It did launch with acarity and by any other standard did well, but the failure to relight could be anything, but I am going for the giant fuel feeder tube bieng the failed part, based on nothing more than how tickled they were with it,and tank baffling bieng a dark art. The slosh of the fuel durring the flip is going to produce an internal tidal wave, lots of stuff gets "tested" there.
Lots of engine failures. Doesn't exactly bode well for a company looking to go public immediately. One of the engine failures was not on the booster but Starship as you noted, and that is a bit unexpected. I don't think they have spoken about it being equal in capability with one engine out, right? Those engines don't move around to compensate IIRC.
Not sure how you come to that conclusion. The capabilities can overcome loss of engines. The fact it was successful with loss of engines shows it is working as designed.
No, it just means the mission happened to be salvageable because of its parameters. The booster is designed to have engines out and can compensate because it has so many engines and many of them are on gimbals. On starship, the vacuum engines aren’t on a gimbal. I’m not sure how it could compensate for one of three engines being out.
Some are on a gimbal and they specifically talked how others gimbaled out a bit to compensate. This is specifically something they designed in and not an accidental lucky save. In this flight they didn’t intend to test “one engine out” feature but it worked out that way.
See my other comment. The vacuum engines are NOT on a gimbal. None of them. The sea level engines on starship and several of the engines on the booster are on a gimbal. But not the vacuum engines for space.
EDIT: I cannot reply further in this thread, but my understanding is that the non vacuum engines are not intended to stay lit throughout the orbital flight in a typical mission. If they are, they can gimbal and compensate.
> The vacuum engines are NOT on a gimbal. None of them
I said some raptor engines are on a gimbal, not vacuum engines.
To be precise, the three central engines can gimbal up to 15 degrees. That can control the thrust vectoring when an engine fails, and that’s what happens in the last flight.
Since the flight already happened and we know it didn’t spin out of control (unless you imply their diagnostic and telemetry was completely off and the engine was actually on) something must have compensated for the failure. It wasn’t magic, it was in fact the central 3 engines that did that.
You may be confused because those are called sea level engines, but that doesn’t mean they can’t work in vacuum.
I think previous comment means "on a gimbal" as in "angled at a non completely prograde direction" (presumably angled such that each engine points through center of mass so that none of the engines impart a torque)
The person you’re replying to is trying to play rhetorical word games.
The upper stage has six engines. The outer three engines are “vacuum engines” (optimized for operation in space). The inner three engines are “non vacuum engines” (optimized for operation in the atmosphere, at sea level).
The outer three vacuum engines are not gimbaled, but the inner three sea level engines are. Thus, it is completely accurate to say that they gimbaled some of the engines to compensate for the engine failure.
They explicitly said that they have engine out capability on the ship in the stream.
That’s for the booster (the big lower part) not for starship (the upper part that continues to space). They were surprised to have a vacuum engine out. In space there’s no atmosphere so you can’t use fins or wings to change direction. And if the engines can’t move around, you only have thrust and gravity and the tiny attitude adjusters to direct your ship.
You're simply wrong. The non-vacuum-optimized engines on the upper stage are still functioning in a vacuum, and their ability to gimbal to offset the loss of one of the vacuum-optimized engines was planned for.
Watch the stream again starting with the ship burn. They explicitly said they have engine out capability on ship. The sea level engines on the ship are running and gimballing.
The takeoff looked almost normal but I noticed a slight drift from vertical, likely because one of the engines was dead or dying. Overall the V3 is supposed to be an upgrade but actual progress is more or less stalling compared V2.
It is supposed to tilt away from the launch tower immediately, you can see this on previous flights. This keeps the engine plume away from the chopsticks and top of the launch tower.
Also an additional goal is to get the booster as far away from the pad as immediately possible in the event it falls back down.
The payload (100t) is at least double that of previous flights. It’s largest spacecraft ever flew. That’s some stalling
There is a slight tilt normally, but I agree it was more than usual.
> The amazing views of the plasma during re-entry, something never seen live before starship, are now routine.
The word "live" is doing a lot of work here. Astronauts used to film the plasma going past the windows of Shuttle.
I remember as a kid my science textbook had a still of it to illustrate plasma.
> The word "live" is doing a lot of work here
A latency of a few seconds for streaming video compared to several months for a still photo from the Shuttle seems an entirely valid use of 'live'.
The best part of this flight was seeing the full reentry with no visible hot spots or burn through like we've seen on every previous reentry of Starship. Seems like they have the heat shields really nailed.
Yes, reuse of the heat shield has been the biggest question mark of the whole program and this is by far the best result of any launch so far. This is the first time it looks plausible that you could consider reusing the heat shield.
I am just delighted that SpaceX continues with the "good enough" pace of development here, at least at these phases. Rapid iteration of build, test, learn, and improve rather than wait for perfection.
They are willing to have "negative outcome learning experiences" to gather data quickly. and, of course, data, data, data.
I like it because I know what insane amount of red tape has built up to do anything similar in a Gov (any Gov).
Shame they're risking that ability with the IPO. We've seen how irrational and ignorant stock traders are from other publicly traded space companies. Even scrubbed launches cause the price to dip.
> We've seen how irrational and ignorant stock traders are from other publicly traded space companies.
Absolutely true, but ignorant stock traders making irrational trades only matters if company management pays attention to them. Musk will maintain complete control of SpaceX even after the IPO, so he can focus on long-term value rather than short-term ups and downs.
Of course, over time, if more shares are issued, this may change.
The argument for the IPO was to fund their space datacenters project, if Musk could afford to ignore the stock price, that wouldn't be needed.
But by the time retail traders can affect the stock price, SpaceX will have already gotten their money.
They get all the money on the ipo day. They dont get more if it goes after unless they sell more stock
The views from Ship's engine bay looked rather ominous -- with the red glow visible in multiple places, and something venting furiously from the broken engine. It was a pleasant surprise that the ship did not explode and not only that, but it even landed exactly on target. Guidance system software engineers have done a very good job!
The booster not completing the return part of the flight was disappointing. They had a similar incident in one of the previous flights, when they tried to maneuver the booster too aggressively immediately after stage separation which caused problems with the fuel supply. If it was something similar this time, it might be solvable by changing just a few details of the maneuver. So, maybe it is not that huge of a deal.
There were many cool things in the webcast, from them showing the catamarans that are deployed at the landing site, to the views form the cameras on-board of the "satellites". The first few minutes after liftoff were just amazing visually.
Hopefully NASA ups their game for Artemis III
My favorite part of this launch that others haven't already mentioned: during reentry, the dummy payload satellites were visible burning up behind the ship!
Glad you mentioned this. I was puzzled by the starry looking background during reentry
This was good forward progress (V3 mostly worked, clear improvements on heat-shield, near-final Starlink deployment system).
Is this enough progress to keep a 2028 crewed landing? Don't know.
I'm curious whether they are going to try to recover a Starship before trying for in-space refueling (or the reverse). Either way, I think both have to work before they can try for an uncrewed lunar landing (presumably in 2027).
The big question is re-usability. How close are they to relaunching a Starship? They may not know for sure until they can get one back intact. If they can launch at least once a month, maybe they'll make it.
If they can re-fly a Starship this year AND demonstrate in-space refueling, then 2027 can be all about an uncrewed landing attempt. That would make me feel good about a 2028 crewed landing on the moon.
> If they can re-fly a Starship this year AND demonstrate in-space refueling
I'd bet that they'll not try in-space refueling before they demonstrated in-space relight of an engine. So they need to fly at least twice. Or even thrice because to demonstrate refueling you need two Starships in orbit.
Seeing both the Starlink mass simulators deploy and the camera view from the last simulators looking back at Starship was really cool.
The amount of data they must have at this point running so many of those raptor engines has got to be insane... at least 300+ engine launches now -- wow.
Five years ago SpaceX reported that they had 30000 seconds of test firing time on the Raptor, over 567 engine starts. Since them the program accelerated dramatically. Well over one thousand engines had been produced, and on an average day at McGregor test facility the Raptors are fired for about 600 seconds. That would give about a million seconds over five years. That's a lot for any engine development program.
Consider for a moment the data requirements for the telemetry system that records those engine runs.
If there's any public info about this I'd love to read it.
Both SpaceX and NASA use LabView. NASA has a relatively detailed description of the engine test stands at Stannis:
https://scholar.google.com/scholar?q=NASA+Data+Acquisition+S...
https://scholar.google.com/scholar?q=Design+of+Electrical+Sy...
A typical test stand would have maybe a thousand channels of relatively slow data (pressures, temperatures, flow rates, valve states, etc), and maybe up to a few hundred of channels for essentially audio data from vibration sensors. This amounts to sub-gigabit per second data rate overall.
If very high speed video / multiple video cameras are used, this could generate massive data rates, but unless something interesting happens it is not clear how important this data is.
In flight, the telemetry data rate from the entire Falcon-9 used to be measured in megabits per second per stage, plus the video stream. It was not a huge amount of data. Presumably now with Starlink they send a lot more telemetry from Starship, but in flight the engines typically have far, far fewer sensors compared to the ground testing.
Very cool. Thank you!
Sort of... this was version 3 of the engine, a fairly big redesign and for version 3 this was the first flight.
Some footage: https://youtu.be/CiWX1nsvqBs?si=lE5autC2y2b8ez2X
At a minute in you can see the satellites being ejected out one by one.
It lifts off so rapidly, it’s truly incredible
“Like a rocket” is a phrase for a reason
Eh, they've left some performance on the table:
> Sprint accelerated at 100 g, reaching a speed of Mach 10 (12,000 km/h; 7,600 mph) in 5 seconds. Such a high velocity at relatively low altitudes created skin temperatures up to 6,200 °F (3,400 °C), requiring an ablative shield to dissipate the heat. The high temperature caused a plasma to form around the missile, requiring extremely powerful radio signals to reach it for guidance. The missile glowed bright white as it flew.
https://en.wikipedia.org/wiki/Sprint_(missile)
Maybe v4.
I’m really struggling to find a point to this comment
Because you are speechless?
Such accelleration! Much wow!
Does anyone know how important hot staging is?
It seems to give the booster a real kick - what's that do to turbo's and fuel movement?
You've got hot exhaust onto cold cryo fuel tank header?
You've got to carry more mass in terms of protection for the tank?
Is doing MECO and then push and then get 100 yards apart or something before second stage / ship engines kick on a big enough penalty to justify all the extra complexity?
All the velocity you lose in any kind of throttle back is velocity you never get back. It hurts you the entire rest of the flight.
I wonder if the hot separation was supposed to be that hot. Going at mach 5 and doing a quick U turn while there was some weird orange color on the side of the Super Heavy, then (possibly?) losing most engines from it seemed extra chaotic
They had a similar issue with the v2 booster the first time, flipped it so hard it damaged the downcomer and they had to change it up a bit. Seems likely here, that thing thats as big as a building flipped ends pretty quick and is partially filled with liquid.
This is amazing.
Oh man, so glad I stayed up to watch it. Kind of a rough start (but it's the 1st flight w/ new redesign, new engines, etc), had an engine out on both booster and ship, but the views were absolutely worth it. They managed to get the last satellite to connect to starlink and download the footage of the ship in orbit. Even with an engine out, the ship managed to reach orbit, deploy all the satellites, re-enter, flip and soft splash into the ocean, near a buoy! And on top of that we got the drone views of the landing. Fucking spectacular views.
It really was an amazing sight to see.
I'm guessing / hoping that the engine outs we're planned, or that they ran the engines with slightly different parameters to test them. If it's just unreliability then it might be a hard problem to solve.
> If it's just unreliability then it might be a hard problem to solve.
It might, but it certainly helps having a ton of them around. Given that they used 42 of them today and 2 failed in some fashion, we'll call that a 1:21 failure rate. On a more typical rocket with say 10 engines (eg falcon 9), there's a good chance they wouldn't have seen the same failure till flight 3.
> Given that they used 42 of them today
20+10+3=33 on the booster, 3+3=6 on the Ship, total 39.
I remember Elon said they want to add 2 engines to the first stage, but that still would be 41. Where's the 42th supposed to be?
I messed up, for some reason I had it in my head that there were 9 on starship, so 33 + 9 = 42.
It’s something like up to 6 can fail and it keeps going, seems pretty good. I know they did some stuff like remove a heat tile to get failure feedback, wonder if engine was planned or accidental
Accidental since they didn’t make the sub-orbit they were aiming for and thus couldn’t test engine re-light.
Very first flight of a brand new engine type (Raptor 3) with totally reworked heatshielding/plumbing/sensors/control systems/etc.
Which is true, but at the same time: this is Starship Flight 12.
The whole point of Starship is that it's a reusable vehicle with easy turnaround and quick maintenance. And in particular it's supposed to be different than the other reusable vehicle with easy turnaround and quick maintenance, which turned out to be sort of a boondoggle.
Yet, they've now hand-built and destroyed twelve of these things across multiple redesigns, and it still hasn't completed its design mission once. In fact basically every launch has unexpected major failures.
As poor as its safety record ultimately ended up being, the shuttle launched successfully on its very first try. And we only had to hand-build five of them. And lost two, sure, which is still a lot less than twelve.
Yes yes, I understand that iterative design has merits and that the ability to rapidly prototype and try things in the stratosphere allows for less conservative tolerances and better ultimate performance.
But does it really take 13+ tries?! At what point to we start wondering if we have another boondoggle on our hands?
If you can afford it, I'm sure anyone developing a rocket would prefer to do it this iterative way. I don't really understand the complain.
The part that makes no sense to me is why they are going starship scale rather than falcon 9 scale. Had they done their prototyping on a rocket with 9 engines on the first stage and 1 on the second, they could have gotten to raptor 3 (and a falcon 9 replacement) while blowing up way fewer engines, launch complexes, etc. There's a reason Spacex started with the falcon 1 rather than the falcon 9. It's a lot cheaper to blow up fewer engines and smaller rockets while you're developing a new rocket engine.
Time costs money as well. Im guessing the cost is irrelevant compared to a faster timeliness.
The point was more that there is a point where (to borrow the software terminology) "iterative design" becomes "death march". Trying a few times in the early days and being willing to throw stuff out and start over is a powerful tool.
I think blowing up a handful of rockets is a fine idea. But at some point you have to ask yourself if it will ever work? Why are we on a another engine redesign? Why is this the third iteration of the second stage? How many more?
And what number is that point? Six? Nine? I'm thinking thirteen may be getting into the danger zone.
In a somewhat similar situation Sergey Korolyov stopped his colleague in front of the Party officials asking a similar question and explained: "We are exploring terra incognita, this is the process of getting knowledge". He was sort of right - even though there were many specific engineering problems, and many of those were rather solvable, especially in hindsight, overall process was stepping into the unknown.
Here we have a cutting edge rocket design - scale, sophistication of engines, design goals - and a commercial evaluation, which path would get to the intended success cheaper. NASA doesn't like public embarrassments, and, as Henry Spencer reminds us, when failure is not an option, the success could be quite costly. So NASA spends billions and many years for a fragile system. If the goal is an airline-like operations, the design should be thoroughly shaken up. It's known that no simulation, no static testing can equate the actual flights in the ability to get the data best describing what conditions the system will encounter in real use. And also, given the industrial scale of Starship production, each flight hardware costs way less than if we'd built them manually, in quantities justifying naming each unit separately.
In Soviet Union, where rocket departments were part of artillery, the testing with actual launches seemed logical. In this case the approach to run a massive test flight program seems logical too, and we can't complain about the lack of progress - first Starship had way less capabilities and performed way worse. In USA we had more than 1000 tests for injector head for F-1 engine in Apollo program, and this number was justified at that time. Starship is way bigger - but the progress is also undeniable, and it would be odd to stop test flights now, when the 3rd iteration of design looks promising.
So, while we can't pin a particular number of tests, I don't think we should worry yet. This year and the next one should be important for Starship program, given SpaceX commitments to help NASA Artemis. If we won't have orbital Starship then - we can come back to this question.
So what if they blow up literally 100 rockets, if they can eventually perfect it faster and more cheaply than the traditional approach, recently typified by SLS.
SpaceX have already proven that the iterative approach works with Falcon 9, literally the most successful rocket program ever. SpaceX have also proven that this specific Super Heavy/Starship rocket design isn’t a dead end. Criticising them for failing to succeed in the future is a valid but uninteresting opinion.
>Why are we on a another engine redesign?
Just looking at it should tell you a lot about why:
https://www.metal-am.com/wp-content/uploads/sites/4/2024/08/...
It’s cheaper and faster to make in volume. It doesn’t require nearly as much shielding, because it’s less fragile, which saves a lot of weight. The engine itself is lighter. And on top of that, it develops more thrust, at higher fuel efficiency.
The net result is cheaper and lifts significantly more mass to space, which significantly drops the cost per kg to orbit.
It already worked, they’re making it much better, and getting it ready for a level of mass production that we’ve never seen anything close to in the space industry, even from SpaceX. They are much more ambitious than I think people who haven’t been watching them closely understand. The US grid is 1.4 TW of generation, they’re aiming to put up 1 TW of AI compute every year. Maybe they’ll stop well short of that, but their stated goal is insanely ambitious.
v3 is the first version that was made with the intention of being used for actual payload delivery. The versions before were about testing and proof of concept.
Reminds me of the adage - The successful have failed more times than the unsuccessful have tried.
But all of those 12 launches happened in just 3 years, and cost a tiny fraction of other major spaceflight development programs.
For reference, SLS has been in development for 5 times as long, and cost 15-20 times as much, as Starship, and they still haven’t landed people on the Moon, which has been one of the stated goals since the Constellation program in 2005.
I don’t see how the number of failures matters if the end result still happens faster and cheaper than anything else.
Recent SpaceX IPO filings put that 'tiny fraction' at about 1/3 or 1/2 of SLS. $15B total investment with about $4-5B of that figure from US gov. Is starship more than 1/2 or 1/3 of the way to a human rated Artemis II style mission? The main reason starship costs less to test (apart from the SLS jobs program baggage) is because of design choices which prevent it from performing such a mission without significant further tech development.
'5 times as long' is dubious too. SpaceX claims to have been working on the design since 2012 vs 2011 for SLS. Ultimately though the start date of a complex program is not well defined, as early conceptual design stages can take years without leaving the drawing board. Government needs to put a start date on such efforts for legal/budget reasons, but a private company does not.
Also relevant - SpaceX has been given a lot of tech and expertise from NASA at a tiny fraction of the cost and time it would have required them to develop it themselves. Therefore, the costs of NASA programs like space shuttle actually includes some of the development costs of SpaceX.
Both programs pale in comparison to Saturn V, which was faster, cheaper, and more technically demanding at the time.
Moreover the two lost shuttles included human lives. Better to blow stuff up with demo payloads now before sending up large contracted payloads or worse human beings!
I couldn’t believe my ears when I first heard that the second ever flight of SLS was going to be crewed.
It worked out in the end, but I can’t imagine being so confident in a new system, no matter how much money and brainpower has been spent to make it safe.
> Better to [...]
That's undeniably true. Nonetheless "Better than the shuttle, which sucked" isn't the design goal.
The question is not even just "is it better to blow up 12 Starships?", which would probably still be true. It's "Why isn't Starship working yet?" and the implied "Maybe Starship sucks too?!".
I don’t keep up with them. What’s different compared to v2?
Now the flaps don't melt! The tiles don't fall off!
It's a major overhaul of the design they've been working on for a long time. There was talk of v3 fixing the problems in early v2 test flights. The booster is v3 as well which presumably is why they had some problems. I believe this is also the first time they flew the v3 engines with the plumbing fully integrated in a single piece housing they 3D printed.
Quite a bit has changed. Here's the highlights: https://www.spacex.com/updates#starship-v3
Big takeaway for me is that the reentry and “landing” of Ship looked great. For the first time, it felt like they’re really on the path to achieving upper stage reuse. That was always the biggest “reach” of the entire program in my view, and today they took a major step forward.
Is it disappointing that they had a couple of engine outs, and also trouble with the booster relight? Sure. Do I have even a little doubt by now that they can fix these problems? None whatsoever.
The success of Ship 39 today was a big, big deal.
Having a faultless payload deploy and a pinpoint landing after losing a whole vacuum engine (one of 3) so early was an unexpectedly amazing performance. I suppose they gimballed the inner non-vac engines to the max and burned longer, next level adaptability.
Most obvious improvement was having no re-entry heating problems, secondmost was deploying with zero issues and with a faster pace. It appears they decided to pause the "horizontal" movement of the pez dispenser before a final push away, probably to avoid vibration causing those "bonks" on the payload door, like we had once before.
What a time to be alive
Cool splashdown.
https://www.youtube.com/watch?v=CiWX1nsvqBs&t=173s
So crazy that the NBC video didn’t include that! It’s still not as bad as NASA skipping the good view of the live separation.
So when's that self-growing Moon base happening?
It's worth remembering that, according to SpaceX's own filings, they've spent >$15 billion on the Starship program thus far with more to come. And SpaceX is burning cash still, particularly because Elon Musk bailed out his own bad decisions with Twitter and xAI with SpaceX stock, basically.
Flight 12 was a relative success. Some engines failed to light but that's an unintended good test. Rockets are typically designed such that they can have a certain number of engines fail and still achieve their mission.
At this point, the entire SpaceX project is a bet on telecommunications services, specifically direct-to-satellite handheld Internet. That's the only market that will recoup the program costs.
We don't have exact figures for the current true cost of a Falcon 9 launch factoring in reuse but many think it's somewhere betweenm $10 and $20 million. Well, SpaceX has spent 100 F9 launches on Starship so far and that's how you have to look at it. Say F9 is $20M and Starship once it starts launching Starlink is $10M that's 150-300+ launches just to break even.
You might be tempted to say there are other missions for Starship but there really aren't. Satellites aren't that bug, as evidences by there being ~1 Falcon Heavy launch per year (usually for the military and/or to geostationary orbit AFAICT). You can't economically put multiple payloads in one Starship because they all have different orbital parameters.
F9 is rated for human spaceflight. It's a long road for Starship to be certified for human spaceflight. SpaceX hasn't even begun to test in-orbit refuelling yet. Gases are weird in microgravity.
F9 is the cash cow funding all this and that too might go away if Blue Origin or one of the other wannabes ever gets a reusable launch platform to commercial operation.
There are big launches like interplanetary missions but those are few and far between.
It would be fascinating if what ends up dooming SpaceX is actually Twitter.
> At this point, the entire SpaceX project is a bet on telecommunications services, specifically direct-to-satellite handheld Internet. That's the only market that will recoup the program costs.
There's also a military angle here. I'll leave it as an exercise for the reader to look into Musk's history with Michael D. Griffin from the Reagan SDI/'Star Wars' program.
It’s an awkward comparison, but F9 can deliver a payload to orbit at a slightly lower price per kg than a Tomahawk missile can deliver it to a target. Starship would be MUCH cheaper if the economics works out the way that SpaceX would like it to.
Obviously a few hundred kg of payload in orbit are not equivalent to the same payload delivered directly to a target.
You don’t need very many kg delivered to target if it’s plutonium. The SDI program had the idea was that if you parked enough defensive weaponry in orbit then maybe mutually assured destruction wasn’t something you had to worry about. The only problem was that getting all that mass into orbit was prohibitively expensive.
Then the deputy director of the program met a young man named Elon Musk, and the rest is history.
> At this point, the entire SpaceX project is a bet on telecommunications services, specifically direct-to-satellite handheld Internet. That's the only market that will recoup the program costs.
I seriously doubt that. Just for example, mining a single asteroid has the potential to flood the market for any number of metals. I don't pretend to know how expensive it would be to achieve that in practice; my point is that there are quite a few different ways to recoup program costs at some handwavey point in the future.
If there were infinite gold bars just sitting on the surface of the moon, it wouldn't be economical to go collect them and bring them back to Earth. No matter how expensive you think any metals are here on Earth, the cost of launching vehicles, rendezvousing with said metals and bringing them back to Earth makes it uneconomical.
An asteroid is much, much further than that but more important than distance is the delta-V required for change its orbit to reach an Earth orbit. So you not only need to get there, which, as discussed, requires in-orbit refuelling with Starship (or any vehicle), but you have to carry all the fuel you need for the orbital burn to bring it back. The rocket equation just kills this immediately.
You really hope you have to get incredibly lucky that an metallic asteroid is on a near-intercept course with Earth that is just shy or going into orbit. The odds for that are, well, astronomical.
That depends on how much a unit of delta-v costs. If you can do the whole mission for $100/kg, quite a few things become economical.
Revenue from xai renting to anthropic this year alone will be more than starlink and launch revenue
Revenue from selling money at a discount isn't generally considered a good strategy.
xAI is burning through $1 billion a month [1]. With Anthropic as a customer, it's basically an argument that we're losing money on every transaction but we'll make it up in volume.
[1]: https://www.bloomberg.com/news/articles/2025-06-17/musk-s-xa...
> Say F9 is $20M and Starship once it starts launching Starlink is $10M that's 150-300+ launches just to break even.
Assuming they deliver the same payload, sure, but that’s very much not the plan.
> That's the only market that will recoup the program costs.
No. If it is just $15B I can think of dozens different usecases ranging from military applications(fast transportation, it is the cheapest ICBM) to asteroid deflection to moon mining to science applications to space datacenter.
Are you seriously thinking $15B is big? Artemis by comparison has spent $93B and has cost of $4B per launch.
Another flight with many explosions and a trivial payload. Trial and error, trial and error. At least these million monkeys have upgraded from typewriters to something more fun.
You don't see the progress between flights, do you?
Some of his employees have died in the meantime, but that's a price Elon is more than willing to pay
Had to check this one and you're indeed right: one death in 2014[0], and one death earlier about a week ago (May 2026)[1] (as well as 600+ workplace injuries [0]).
[0]: https://www.reuters.com/investigates/special-report/spacex-m...
[1]: https://techcrunch.com/2026/05/18/osha-probing-worker-death-...
Scaled Composites has had a few deaths, too.
Less recent that I thought: https://spacenews.com/explosion-mojave-air-and-space-port-ki...
How many deaths has Facebook or Google had in that time?
Over 100 died building the hoover dam. Over 5000 building the Panama canal, and over 30,000 if you count the failed attempt.
30,000 people trip and die in their homes per year.
70,000 Americans die per year from medication error or accidental overdose.
This incremental progress, far smaller improvements than planned, has put them so far behind schedule I'm not confident this design is any good. Still haven't done orbit. This launch was not a smooth launch. SLS by contrast seems to work. Why did nasa contract SpaceX for the lander. The whole plan is bad.
Are you serious? They just launched a completely revamped version of Starship from an entirely new pad, and still hit almost all of their planned milestones while demonstrating that the design is reliable enough to handle a missing engine.
Go back and look at the original plans and projections. Constantly redesigning is not something to be proud of. I call it vibe spaceship design.
Two successful test missions over 15 years using a dead end rocket and ship design for $50+ billion is not something to be proud of either.
Starship's design is so far beyond where the rest of the world is that even if it takes another 3 years of iteration to perfect it will still be by far the best rocket in the world for many years afterward, to the point where it would hardly make sense to choose to launch on any other existing or currently in development rocket for any reason other than political ones.
There's plenty of finicky systems which go on to be good systems with a lot of work. Some things are just hard, a lot of the time you just don't see them being hard so publicly.
Sometimes a design is poor and needs a lot of modification and patches to kind of work.
That's software engineer "I'll simplify it!" logic and the score of it's successes is far too low for how common it is.
If a design with a bunch of modifications works, then it's a good design. Thinking you need to clean sheet redesign everything is how you get Second System Syndrome.
Nerd bait. Humans belong on earth.
Columbus belonged in Italy.