Errant upper stage spoils Blue Origin’s success in reusing New Glenn booster

Blue Origin's reused first stage hit its targets, but New Glenn's upper stage did not.

See full article...

 

[wagnerrp]

Yeah it is very atypical. For max sats no they couldn't afford to do it this way. After first post I went back and did that math. The plane change would cost a staggering 1,700 m/s so normally would be something to avoid BUT with 6t vs 36t capability the upper stage had dV to spare so maybe they did intend that to booster the booster odds. Hard to say. Would have been nice if BO shared more details prior to launch. If they had we could go "yeah everything was nominal at SECO1" so the failure was post SECO1 and it looks like a simple relight failure. If the target at SECO1 was 49 deg well 36 is worse than the relight failure.

On edit: someone on NSF brought up an alternate theory. The rocket was off course (49 vs 36 deg inclination) as such they intentionally didn't make the SES2 burn so the low apogee is the symptom not the problem. Again with BO sharing so little data before launch on how this launch was supposed to look hard to say.

They wouldn't do a plane shift after reaching orbit unless they were choosing to show off. They would dogleg immediately after separation, when they're only going 2-2.5km/s, yielding far reduced requirements.

 

[danan]

Can you imagine the swing from elation on a successful launch and ascent to the crushing blow finding out there is an upper stage failure and your company will be writing a nine figure check.

My understanding is that insurance actuaries are very good at what they do. I expect they’ll have spread the risk, hedged it, accounted for this being early stages of NG, and probably priced in at least one failure at some point. They are the “house” in this type of gambling, and they stack the deck as much as they can in their favor.

Also, writing sometimes big checks to cover claims is one of the things insurance companies do. Probably without even blinking. The trick is to cash checks worth a lot more than the ones you’re writing.

 

[SirOmega]

Will it though.

Launches are usually insured in advance. Launches for the next year probably two years are already locked down. So any new insurance policy would be for launch 10 or 15 not launch 5. The insurance executive for launch 5 likely has hearburn right now but that contract was signed a long time ago.

If BO nails the next 10 launches insurance prices might go down not up. Also I don't think anyone in the business thinks it looks "much" riskier. They were likely assumming a failure rate of at least 5% possibly 10% given lack of history and 1 failure in the first 5 doesn't mean that assumption is incorrect. Now if they lose another payload in the next couple that changes thing.

The key to this discussion is about the losses here.

Based on what little I can find for the cost of the BB7 satellite lost today, it isn't a major insurance catastrophe. In 2023, when the satellite insurance industry appeared to lose more than they took in on premiums, that's when costs increased meaningfully.

If BB7 only cost around 30M, that is not a huge hit, compared to Viasat 3-1 and Immersat F6-2 which were each 300M+ claims.

 

[fl4Ksh]

That AST satellite costs $22M. I'm sure that New Glenn has larger, more expensive payloads on its launch schedule. Space Force likely is gearing up for an in-depth post flight investigation before putting one of their $500 million(+) military satellites on NG.

 

[Statistical]

The key to this discussion is about the losses here.

Based on what little I can find for the cost of the BB7 satellite lost today, it isn't a major insurance catastrophe. In 2023, when the satellite insurance industry appeared to lose more than they took in on premiums, that's when costs increased meaningfully.

If BB7 only cost around 30M, that is not a huge hit, compared to Viasat 3-1 and Immersat F6-2 which were each 300M+ claims.

Yeah better than Viasat but launch insurance usually covers the launch too. What is the price of a NG? We have AFAIK only one data point from Amazon filings and that is $120M. Then again AST is going earlier, on the first reflown booster, and willing to accept a hotfire with payload so maybe BO cut them a sweet deal for their first launch.

 

[wagnerrp]

From the article;

”Officials reported a similarly precise trajectory on the second New Glenn launch, which dispatched two NASA science probes toward Mars in November.”

I keep in mind that New Glenn has successfully lunched two probes sent to Mars on only its second flight. Also this recent NG launch is only its third flight. IMO it will take more time/flights to accurately decide on how viable NG is as a launcher.

A distinction without a real difference, New Glenn launched two probes to a weakly bound Earth orbit. Those probes will spend ~1.5km/s of their own propellant to make their own way to Mars at the next transfer window in November.

 

[Rachelhikes]

That is a false opposition. The mission is both getting the booster back and delivering the payload.

if you don’t deliver the payloads, you won’t get any more. Especially not high-value payloads. SpaceX delivered plenty of payloads while working on booster re-use. Destroying payloads while recovering boosters is not a business at all.

 

[BadSuperblock]

What were the odds of failure?

Uh, can we please keep in mind the name of…

…this booster, named Never Tell Me The Odds

 

[danan]

There is no dozens. It was 14 to 16 and F9 is a total of 7. The question is if it were you would you put 8 sats on the next NG launch after losing one on this launch. If not then that 8 sat launch becomes 1. So the company is looking at 8 sats deployed by end of year vs planned 16.

If I’m ramping my satellite manufacturing rate to the point it exceeds the expected launch rate, I might risk more than one. If time is the largest cost, then what matters is how much more time is being lost by another failure vs how much might be gained by putting more birds up with a success.

 

[Wickwick]

My understanding is that insurance actuaries are very good at what they do. I expect they’ll have spread the risk, hedged it, accounted for this being early stages of NG, and probably priced in at least one failure at some point. They are the “house” in this type of gambling, and they stack the deck as much as they can in their favor.

Also, writing sometimes big checks to cover claims is one of the things insurance companies do. Probably without even blinking. The trick is to cash checks worth a lot more than the ones you’re writing.

Reinsurance is how insurance companies handle that.

 

[Wickwick]

A distinction without a real difference, New Glenn launched two probes to a weakly bound Earth orbit. Those probes will spend ~1.5km/s of their own propellant to make their own way to Mars at the next transfer window in November.

True, just about any near-escape trajectory probably could have worked. However, it does appear Blue hit the targeted trajectory for that mission.

 

[Statistical]

I have to agree with Eric on this one.


View: https://x.com/SciGuySpace/status/2045954872957170063


The AST yolo 100% of their IRA into a speculative stock WSB style stockbros are beyond obnoxious. They haven't found ars but their low information bleeting and ranting is bringing down signal to noise information everywhere space nerds are discussing the event.

 

[InvariantCapitalist]

Sample size = one so take what I write next with a grain of salt, but I wonder if this is more evidence that the design model used by Falcon 9 is superior. It’s made over 650 launches with only two upper stage failures, so how much of that is due to Merlin having over 6,500 launch ignitions, ie getting further on the quality curve by sharing uses with first stage.

If they had designed F9 to use a different engine on the second stage they’d only have 10% of the uses, and performance data.

 

[RegisteredUser]

Re: Insurance

SpaceNews, in their article covering BO's upper stage failure, had a link to a recent SEC filing by AST SpaceMobile in which details regarding insurance claims, premiums, coverages, and overall risk management, including time delays, involved with launching their payloads are listed in some detail.

The March 2, 2026 SEC relevant insurance information can be found on page 23:

Launch insurance, even if it is available, will not fully cover the risks related to the launch of our satellites.

We have and expect to continue to insure the launch of all or a portion of our satellites to operate the SpaceMobile Service as intended, but do not intend to insure our satellites once they are launched for their remaining in-orbit operational lives. Launch insurance currently costs approximately 3.0% to 20.0% of the insured value of the satellite (including launch costs) but will vary depending on market conditions and the safety record of the launch vehicle. In the future, we may choose not to insure every launch or to only partially insure some or all launches. Even if a lost satellite is fully insured, acquiring a replacement satellite may be difficult and time consuming. Furthermore, the insurance does not cover lost revenue.

We expect any launch failure insurance policies that we obtain to include specified exclusions, deductibles and material change limitations. Typically, these insurance policies exclude coverage for damage arising from acts of war, lasers, and other similar potential risks for which exclusions are customary in the industry at the time the policy is written.

If launch insurance rates were to rise substantially, all of the launch costs would increase. Also, in light of increasing costs, the scope of insurance exclusions and limitations on the nature of the losses for which we can obtain insurance, or other business reasons, we may conclude that it does not make business sense to obtain third-party insurance and may decide to pursue other strategies for mitigating the risk of a satellite launch failure, such as obtaining relaunch guaranties from the launch provider. It is also possible that insurance could become unavailable, either generally or for a specific launch vehicle, or that new insurance could be subject to broader exclusions on coverage, in which event we would bear the risk of launch failures.

This may clear things up about how AST SpaceMobile itself assesses risk management involved with launches and costs.

Edit: I tried to post the link to the SEC listing but it was blocked but I'll try again:

https://www.sec.gov/ix?doc=/Archives/edgar/data/0001780312/000178031226000006/asts-20251231.htm

 

[SteveBigandTall]

Look they got the booster back!
And the payload too...in a less than new configuration.

 

[r0twhylr]

Yeah it is very atypical. For max sats no they couldn't afford to do it this way. After first post I went back and did that math. The plane change would cost a staggering 1,700 m/s so normally would be something to avoid BUT with 6t vs 36t capability the upper stage had dV to spare so maybe they did intend that to booster the booster odds. Hard to say. Would have been nice if BO shared more details prior to launch. If they had we could go "yeah everything was nominal at SECO1" so the failure was post SECO1 and it looks like a simple relight failure. If the target at SECO1 was 49 deg well 36 is worse than the relight failure.

On edit: someone on NSF brought up an alternate theory. The rocket was off course (49 vs 36 deg inclination) as such they intentionally didn't make the SES2 burn so the low apogee is the symptom not the problem. Again with BO sharing so little data before launch on how this launch was supposed to look hard to say. That is made harder by very vague statements of BO. We put the sat into wrong orbit doesn't necessarily mean a relight error just what many (myself included) assumed.

At least in that case they wouldn't need to investigate and fix whatever was wrong with the relight, and might be flying again sooner rather than later.

But how does a launch provider launch to the wrong inclination? And wouldn't that be immediately apparent? Would it be possible to correct course at that point?

 

[GFKBill]

Re: Insurance

SpaceNews, in their article covering BO's upper stage failure, had a link to a recent SEC filing by AST SpaceMobile in which details regarding insurance claims, premiums, coverages, and overall risk management, including time delays, involved with launching their payloads are listed in some detail.

Thanks! That clears up a thought I was about to post - I was surprised it was cost-efficient for the satellite owner to insure at all when they're planning a constellation. Sounds like once (if) they get to a good cadence and reliable launches, they won't bother to insure.

 

[Dtiffster]

At least in that case they wouldn't need to investigate and fix whatever was wrong with the relight, and might be flying again sooner rather than later.

But how does a launch provider launch to the wrong inclination? And wouldn't that be immediately apparent? Would it be possible to correct course at that point?

There have been instances in the past, the one that spring to mind was an Ariane 5 launch, where the incorrect trajectory was loaded into GNC. Given that they got the booster back the booster obviously flew the correct trajectory, but maybe the upper stage was supposed to dog leg and someone screwed the pooch. The rocket is autonomous during that initial burn, all they could do is trigger the FTS.

 

[tsunam]

The bizarre thing is that lots of mainstream press seem to have missed this (or BO’s press release to them only mentioned the successful landing and they didn’t bother to check the whole story?)
I’ve noticed that articles on both Reuters and the Economist only mention the successful landing and how there’s now viable competition for SpaceX and they don’t appear to realize the launch failed.

I know that mainstream news orgs aren’t the savviest on science/tech, but missing that the satellite didn’t get to where it was supposed to seems like a rather glaring miss..

 

[Reynold]

Re: Insurance

SpaceNews, in their article covering BO's upper stage failure, had a link to a recent SEC filing by AST SpaceMobile in which details regarding insurance claims, premiums, coverages, and overall risk management, including time delays, involved with launching their payloads are listed in some detail.

The March 2, 2026 SEC relevant insurance information can be found on page 23:



This may clear things up about how AST SpaceMobile itself assesses risk management involved with launches and costs.

Edit: I tried to post the link to the SEC listing but it was blocked but I'll try again:

https://www.sec.gov/ix?doc=/Archives/edgar/data/0001780312/000178031226000006/asts-20251231.htm

I was able to read that link, but would describe that SEC listing as pretty vague; 3-20% of a satellite cost, (which itself could vary between $1M and $500+M, see Viasat or Defense Department birds), is a rather big range. There is also no description of how this insurance cost varies based on risk management or time delays. Mostly boilerplate SEC language of "this could get more or less expensive by some unknown amount".

Which is normal for a SEC filing, but not super helpful for us data nerds.

 

[windnwar]

Not that it is tied to this failure but didn't Blue recently burst a second stage tank during testing in the last month as well?

 

[sxotty]

That seems likely to me. Since the upper stage is likely intact hopefully BO got plenty of good telemetry to figure out exactly why it didn't relight. For ksp fans hopefully they didn't press space instead of z in their excitement.

That is the cool thing about recovering it. Hopefully it is useful to figure the problem out

 

[dmsilev]

That is the cool thing about recovering it. Hopefully it is useful to figure the problem out

They don’t recover the second stage. But, since whatever the failure was, it didn’t include an explosion, they had plenty of time to download as much sensor data as they want, maybe even try another relight after separation from the payload.

 

[Cloudgazer]

That is the cool thing about recovering it. Hopefully it is useful to figure the problem out

they're not recovering the stage which failed

 

[Bannerdog]

The mission isn't getting the booster back, the mission is delivering the payload to where it is supposed to go. Customers don't care about the launch company's booster — they care about their payload. Blue Origin needs to prove it can do the part that matters reliably. Pretty damn disappointing, even with a successful booster landing.

Obviously, what the customer cares about is whether their payload is successfully delivered.

However, the ability to reuse the first stage will greatly decrease the launch cost paid by a customer (which the customer certainly cares about).

And, this is only the third launch of New Glenn. Reliability will almost certainly improve.

Was this launch disappointing from the customer's perspective?

Of course.

However, for the long term, the successful first stage landing overshadows (IMO), the second stage issue.

In the long run, that will also be true even for the customer losing a satellite, provided that BO can avoid future failures (or, have a very small percentage of failed missions).

In the long run, cost savings can more than make up for the loss of a satellite.

I would not characterize the mission as a disappointment (except to THAT customer).

In terms of expectations of lowered launch costs, it was encouraging to see a repeat of a successful landing, suggesting that the first such landing was not an anomaly.

 

[TimeToTilt]

The bizarre thing is that lots of mainstream press seem to have missed this (or BO’s press release to them only mentioned the successful landing and they didn’t bother to check the whole story?)
I’ve noticed that articles on both Reuters and the Economist only mention the successful landing and how there’s now viable competition for SpaceX and they don’t appear to realize the launch failed.

I know that mainstream news orgs aren’t the savviest on science/tech, but missing that the satellite didn’t get to where it was supposed to seems like a rather glaring miss..

Maybe time related depending on when those stories were written. The booster landing happened minutes after launch, the problem with the orbit insertion was hours later.

 

[wagnerrp]

Why would they have launched southbound?

That launches to ~34°, and orbital track placed them at 36.1°. It's almost like they just forgot to dogleg. 49° is well within KSC's azimuth range, so why wouldn't they just launch directly into orbit northbound? Avoiding bad weather or rough seas at the landing site?

 

[dmsilev]

With regards to insurance, I have to imagine that the new-launcher-risk premium was baked into the rate calculations from the get-go. The odds of something going wrong enough to cause total payload failure are nontrivial. If there’s a repetition or a pattern of failures, that will be a problem, but one failure out of the first three flights in and of itself, not so much.

 

[NetMage]

If you cant deliver, the mission has failed for the customer, as simple as that, even if they cut the price by 50%, noone will want to flight if they cannot prove reliable delivery.

As AstraSpace demonstrated for everyone. Satellites aren’t as disposable as they thought.

 

[Dtiffster]

I feel that, without reusing the engines this is partial reuse at best. Reusing some tanks and avionics, while useful, is missing a big part of the point. That’s why ULA was still pretending to target ‘smart reuse’ that only recovered the engines, because they are the most valuable part.

They did recover the engines and are planning on reusing them on another flight. Obviously they would like to get to the point where they don't have to pull and refurb engines before they reuse them, but they don't need to get there immediately. And I think you are majorly underselling the rest of the rocket by saying 'just some tanks and avionics', NGs booster is a beast. It takes them over a year to slap together those tanks and avionics, they definitely need to reuse them as well as the engines.

 

[normally butters]

They don’t recover the second stage. But, since whatever the failure was, it didn’t include an explosion, they had plenty of time to download as much sensor data as they want, maybe even try another relight after separation from the payload.

Launch vehicles are autonomous. Even the flight termination system is autonomous nowadays. There's no commanding from the ground. The upper stage will do whatever it was programmed to do in response to whatever anomaly occurred. Every off-nominal code path in the flight software must lead to a reasonable predefined response, which cannot be altered in flight. The telemetry will include as many data points as they deem useful and can fit in the available stream bandwidth, with no way to query the vehicle for additional data to downlink later.

 

[Statistical]

I feel that, without reusing the engines this is partial reuse at best. Reusing some tanks and avionics, while useful, is missing a big part of the point. That’s why ULA was still pretending to target ‘smart reuse’ that only recovered the engines, because they are the most valuable part.

It is the first reuse. BO says they plan to fly the engines from NG2 on a future LV after extensive testing. It was just faster to use new engines they had.

To put it into perspective SpaceX didn't reuse a booster for 464 days after the first landing. Nobody else has ever reused a booster at all. In the case of SpaceX that first recovered booster (B1019) was never reused at all. The first reused booster was B1021 and it flew 356 days after it was recovered. For the first two years boosters were either reused at most once, lost while landing, intentionally expended, or retired without reuse.

It wasn't until 2018 (over 2 years after first landing) that boosters started having turn around times of <90 days, landing failure rates dropped below 10%, and boosters were reused 2+ times. That was the point where reuse starting having a material impact on cadence and cost. For BO that comparable date would be 2028.

These days SpaceX reuses boosters so often and so quickly it feels like that has always been the case but it took years many years to get to that point. 2021 (>5 years after first landing) was the first time a booster flew 10 times. 2024 (>8 years after first landing) was the first year they had a turnaround time of <30 days.

 

[Juvba Fnakix]

The bizarre thing is that lots of mainstream press seem to have missed this (or BO’s press release to them only mentioned the successful landing and they didn’t bother to check the whole story?)
I’ve noticed that articles on both Reuters and the Economist only mention the successful landing and how there’s now viable competition for SpaceX and they don’t appear to realize the launch failed.

I know that mainstream news orgs aren’t the savviest on science/tech, but missing that the satellite didn’t get to where it was supposed to seems like a rather glaring miss..

Mainstream news places value on providing the latest news. They buy the first story available. The only practical way for a journalist to sell into that market is to write the story first, attend most of the event, possibly make and edit during then rush out before the end to make a sale.

Stephen and Eric have an enormous advantage here because they can wait until the end of the event without risk of losing the sale. The only way mainstream will improve would be for readers to express a preference for late and accurate over latest.

 

[jhodge]

Would it be expected that BO owes AST another launch (or a discount on a launch), or would the insurance cover the launch cost as well as the payload?

 

[wagnerrp]

Would it be expected that BO owes AST another launch (or a discount on a launch), or would the insurance cover the launch cost as well as the payload?

The launch happened as soon as it left the pad. Insurance covers the payload. The circumstances were different with AMOS-6 because it was lost during a test, not during the launch.

 

[ai_fodder]

nice...

I point out a typo that is corrected, silently, then I look like I am either making a bad pun or just a jerk
nice...

 

[brionl]

Would it be expected that BO owes AST another launch (or a discount on a launch), or would the insurance cover the launch cost as well as the payload?

It probably depends on the terms of the contract.

 

[eyesoars]

I was able to read that link, but would describe that SEC listing as pretty vague; 3-20% of a satellite cost, (which itself could vary between $1M and $500+M, see Viasat or Defense Department birds), is a rather big range. There is also no description of how this insurance cost varies based on risk management or time delays. Mostly boilerplate SEC language of "this could get more or less expensive by some unknown amount".

Which is normal for a SEC filing, but not super helpful for us data nerds.

So far as I am aware, all U.S. DOD launches are self-insured. I.e., if DOD loses a satellite on launch, it eats the loss.