Will a Starship go to Mars in the 2026 launch window?

[TheRadicalModerate]

>"Recovery is convenient, but not necessary."

I would suggest reuse of ships is useful to get in more test flights but may not be necessary if everything goes really well.

Things have to go really well for there to be any chance of an interplanetary launch in '26.  If they go really well, then teardowns probably don't tell you as much.  This is an extremely heavily instrumented vehicle, which relies on Starlink to dump massive amounts of data back to the ground.

The key is manufacturing rate of new Ships.  (NB:  I'm assuming booster reuse happens.  Without that, they'll never get the launch cadence they'd need.)

>"Easier just to go straight to depot.  Any ol' Starship will work as a tanker."
"Any random Starship will work."

Easier? You then have 3 ship types to develop Mars, depot and Tanker. I would have thought easier with just two types to develop Mars and Tanker. Either way you have to develop hardware for supplying propellants and receiving propellants so maybe there isn't much in it?

For testing purposes maybe any [new if no reuse] Starship will work as a tanker for testing propellant transfer purposes, but you still need to build multiple tankers for the refuelling campaign as ordinary Starships just won't have fuel capacity to be much use, unless you are designing tanks for payload section.

Let's do some back-of-napkin math to figure out what's available from a vanilla-flavored Ship with no payload.  We'll assume:

dry mass = 130t
prop mains = 1550t
Nominal payload to LEO = 100t

Mass ratio = (130 + 100 + 1550) / (130 + 100) = 7.74. 

That's what's needed to get a Starship from staging to LEO and back to landing.  Now let's run the same Ship with no payload, which should require the same mass ratio:

(130 + 0 + 1550) / (130 + propRemain) = 7.74
propRemain = 87t

So you're incurring a 13% "payload" penalty if you don't stretch the tanks.  (The payload is obviously the prop deliverable to the depot.)  That's a non-trivial penalty, but it's still more on the "optimization" side of the ledger than the "essential" side.

I suspect that number will be swamped by depot boil-off.  Much better to put the work into that system than into stretching the tanks.  (NB:  Stretching the tanks is nice, and fairly easy: you just move the domes, but there's a certain amount of structural analysis that goes into re-kludging the stringers, doublers, etc.  FWIW, one ring segment = 104t of average-density methalox.  Different densities for LCH4 and LOX, obviously.)

In terms of schedule, a vanilla Ship for a tanker is probably the way to go.

Waiting for recovery and a refuelling test before starting to build tankers for the refuelling campaign already seems to be impossible to me if it takes a total of 6-9 months for full build of a ship/tanker.

That number sounds way, way too slow.  The gating item is engine production, and that's reputed to be at >5 Raptor 2's per week.  I'd guess that Raptor 3 will ramp to those rates fairly quickly, which means that they'll be able to produce and engine set in less than 2 weeks.

TPS and flaps are a fairly long-lead operation, but if you've decided to go expendable, those aren't needed operations.  A ship every two weeks should feed two pads with a 4-week turnaround.  That gives SpaceX the ability to launch 15 expendable ships between now and when the interplanetary flight would need to go.

They won't do expendable ships unless they need to--or they're really emphasizing a Mars attempt over getting reusability up and running, commencing Starlink v3 (fka v2 maxi) deployments, and HLS commitments.  That's nuts.

So, while an expendable Mars campaign is theoretically possible if everything goes perfectly, it comes at a very heavy price.  Then, if everything doesn't go perfectly, they're left with no Mars campaign, re-use pushed back, Starlink v3 later than they'd like, and NASA angry with them.

[TheRadicalModerate]

Good deletions, but I see you're riding your old "I need need need comms during blackout" hobby-horse. Obviously this part is a nice-to-have, not really a need-to-have.

That hobby-horse still seems to hold up pretty well, especially if the rationale behind messing up everything on your schedule is to get the engineering data needed to make the '28-'29 window as likely to succeed as possible.  If all you're getting is the seven-minutes-of-terror-style tones, that doesn't really move the ball forward--unless the EDL actually succeeds, which has a near-zero probability of occurring.

I guess they could engineer some kind of black box that can survive a breakup event, parachute to the surface, and transmit the data, but that sounds like roughly the same amount of work as designing a JPL-style lander.  Seems a lot cheaper--and more useful--to tackle the comms blackout problem with tech that they already mostly have.

BTW:  the comms I'm talking about in the list is just vanilla interplanetary comms.  It's not the end of the world to do it, but it's definitely an item if you plan to hit a Mars entry window.  You have to be able to control the vehicle in transit.

[envy887]

Is the least thing more than a flyby, an aero-non-capture? (aiming for some aerocapture data even if it is not ideal data)

If COSPAR probably won't approve a Hail Mary inadequately prepared plan that involves an aerocapture then this might also be the most they could aim for?

Would a long lasting (high impulse?) RCS system be all they needed for this or would they also need (proven?) raptor relight ability and propellants to get the power needed to escape if the aero-non-capture did not work as planned?

Is there anything else that can be added to a flyby mission short of aerocapture? Drop off Marslink sats a couple of months before flyby and use all the power they can generate over that period for propulsive capture efforts?

MSL only needed about 9 m/s in course corrections, which is well within the capability of a cold gas system even for something as large as Starship. I don't think you'd want to use the Raptors for anything besides TMI and landing.

[TheRadicalModerate]

If COSPAR probably won't approve a Hail Mary inadequately prepared plan that involves an aerocapture then this might also be the most they could aim for?

Everything we've talked about requires the Trump administration waiving both Category III and IV planetary protection guidelines, which in turn requires the US to tell COSPAR to go pound sand.  Without such a waiver, the best that can happen is a propulsive capture at Mars, which yields no EDL data whatsoever.

Back when Elon and Trump were buddies, the waiver seemed like a forgone conclusion.  Now... who knows?

This is yet another reason why I'd weight a '26 attempt as less likely.

[sdsds]

[...] Let's do some back-of-napkin math to figure out what's available from a vanilla-flavored Ship with no payload.  We'll assume:

dry mass = 130t
prop mains = 1550t
Nominal payload to LEO = 100t

Mass ratio = (130 + 100 + 1550) / (130 + 100) = 7.74. 

That's what's needed to get a Starship from staging to LEO and back to landing.  Now let's run the same Ship with no payload, which should require the same mass ratio:

(130 + 0 + 1550) / (130 + propRemain) = 7.74
propRemain = 87t

I'm confused. This ship can deliver 100t of payload to LEO, unless that payload happens to be prop, in which case it can deliver only 87t?

[crandles57]

Waiting for recovery and a refuelling test before starting to build tankers for the refuelling campaign already seems to be impossible to me if it takes a total of 6-9 months for full build of a ship/tanker.

That number sounds way, way too slow.  The gating item is engine production, and that's reputed to be at >5 Raptor 2's per week.  I'd guess that Raptor 3 will ramp to those rates fairly quickly, which means that they'll be able to produce and engine set in less than 2 weeks.

TPS and flaps are a fairly long-lead operation, but if you've decided to go expendable, those aren't needed operations.  A ship every two weeks should feed two pads with a 4-week turnaround.  That gives SpaceX the ability to launch 15 expendable ships between now and when the interplanetary flight would need to go.

They won't do expendable ships unless they need to--or they're really emphasizing a Mars attempt over getting reusability up and running, commencing Starlink v3 (fka v2 maxi) deployments, and HLS commitments.  That's nuts.

So, while an expendable Mars campaign is theoretically possible if everything goes perfectly, it comes at a very heavy price.  Then, if everything doesn't go perfectly, they're left with no Mars campaign, re-use pushed back, Starlink v3 later than they'd like, and NASA angry with them.

Thank you again.

I agree that you can get a finished ship off the end of the line every 2-4 weeks or so. However this isn't the full time for building a ship. If you have 6 ships in production to get a 3 week rate then that is more like 18 weeks for the full build time. At present we have what? 37-46 in various stages of production so maybe nearer 30 weeks in total which fits neatly in my 6-9 months. Also ship 38 parts first seen early December 2024 and it might get launched in September which might be nearer 10 months but they can probably be more efficient when not reworking stuff and pausing due to mishap investigations.

But since you have shown they can use any ship as a tanker just add equipment for supplying propellants to the mars ship, that long period no longer seems so relevant. You do need the ships to be used to be in production at least 6 months before the window closes and a few boosters if you haven't got rapid reuse. Hopefully they do have time to get the booster reuse turnaround time down if not also ship reuse.

Committing to building ships and boosters for Mars attempt instead of ships for testing purposes could well be problematic for delaying Starlink deployment commencement and for NASA as you indicated. They might need to make that commitment decision as soon as ~April 2026? If they have demonstrated ship catch and refuelling by then maybe then NASA would be impressed enough by that and think an attempt at a small refuelling campaign would seem a worthy attempt along road to HLS demo? It isn't much time to get that done though.

[crandles57]

[...] Let's do some back-of-napkin math to figure out what's available from a vanilla-flavored Ship with no payload.  We'll assume:

dry mass = 130t
prop mains = 1550t
Nominal payload to LEO = 100t

Mass ratio = (130 + 100 + 1550) / (130 + 100) = 7.74. 

That's what's needed to get a Starship from staging to LEO and back to landing.  Now let's run the same Ship with no payload, which should require the same mass ratio:

(130 + 0 + 1550) / (130 + propRemain) = 7.74
propRemain = 87t

I'm confused. This ship can deliver 100t of payload to LEO, unless that payload happens to be prop, in which case it can deliver only 87t?

It is 1550 ton of propellant with either of these, the tanks are only so large. By not having 100 tons of payload in payload section you save 87 tons of fuel.

If you had 100ton of propellants in the payload section then it would be 100tons but then it would be 1650 tons of propellants to get 100 tons of propellant to orbit. (As opposed to 1550 to get 87)

Hopefully you see the difference now?

[Twark_Main]

Good deletions, but I see you're riding your old "I need need need comms during blackout" hobby-horse. Obviously this part is a nice-to-have, not really a need-to-have.

That hobby-horse still seems to hold up pretty well

Nothing has happened in the news or announcements lately for it to "hold up" against, so this is just you bragging that you still believe your own theory. 

especially if the rationale behind messing up everything on your schedule is to get the engineering data needed to make the '28-'29 window as likely to succeed as possible.  If all you're getting is the seven-minutes-of-terror-style tones, that doesn't really move the ball forward--unless the EDL actually succeeds, which has a near-zero probability of occurring.

I guess they could engineer some kind of black box that can survive a breakup event, parachute to the surface, and transmit the data, but that sounds like roughly the same amount of work as designing a JPL-style lander.  Seems a lot cheaper--and more useful--to tackle the comms blackout problem with tech that they already mostly have.

BTW:  the comms I'm talking about in the list is just vanilla interplanetary comms.  It's not the end of the world to do it, but it's definitely an item if you plan to hit a Mars entry window.  You have to be able to control the vehicle in transit.

"Hypersonic reentry telemetry" is not a feature of "vanilla interplanetary comms."

Tech we mostly already have is using existing orbiters as relays, and existing imaging assets to monitor the reentry.

Even just knowing the exact time of breakup and the probable breakup mode (how many pieces) would be a hugely valuable data point, and way more information than most Mars landers ever had access to.


(Also if you did have a black box, you'd just transmit the data after comm blackout and before it impacts the ground. No parachute, no landing, simple batteries instead of solar panels, high power omni transmitting to nearby orbiters, much much much simpler than a lander)

[TheRadicalModerate]

[...] Let's do some back-of-napkin math to figure out what's available from a vanilla-flavored Ship with no payload.  We'll assume:

dry mass = 130t
prop mains = 1550t
Nominal payload to LEO = 100t

Mass ratio = (130 + 100 + 1550) / (130 + 100) = 7.74. 

That's what's needed to get a Starship from staging to LEO and back to landing.  Now let's run the same Ship with no payload, which should require the same mass ratio:

(130 + 0 + 1550) / (130 + propRemain) = 7.74
propRemain = 87t

I'm confused. This ship can deliver 100t of payload to LEO, unless that payload happens to be prop, in which case it can deliver only 87t?

This does assume that total delta-v is the same for both cases.  Gravity losses should be slightly less in the no-payload case, which will help a bit.  It's hard to estimate, but I'd be surprised if the total delta-v varied by more than 1%.  That would make the mass ratio only 7.58.  So:

(130 + 0 + 1550) / (130 + propRemain) = 7.58
propRemain = 91.6t

So only 8.4% difference.

Note:  If you made the tanks 100t bigger, then the "payload" would indeed be exactly 100t.  The intuitive way to think about this is that you have less prop to burn, because you're carrying the payload in the tank, instead of in the payload bay.

[TheRadicalModerate]

"Hypersonic reentry telemetry" is not a feature of "vanilla interplanetary comms."

Hey, you're the one who wanted to insult my hobby horse.  I was only talking about the vanilla stuff.  It's not a huge deal, but it is a deal.  So's the nav stuff.  It's not like you're reinventing the wheel, but it's a non-trivial task.  That's why it belongs on the list.

Tech we mostly already have is using existing orbiters as relays, and existing imaging assets to monitor the reentry.

That's fine if you happen to have an orbiter that you can send to through the plasma sheath.  The thing we'd discussed before (your "trailer" idea) solves that by being more-or-less behind the entering ship, on more-or-less the same trajectory.

Even just knowing the exact time of breakup and the probable breakup mode (how many pieces) would be a hugely valuable data point, and way more information than most Mars landers ever had access to.

No, it's exactly the same info they generate via the tones.  One can quibble about how many tones you have, but JPL is pretty confident in their current lander architecture.  Starship, not so much.

The big unknown for Mars is how the TPS is gonna oxidize.  I imagine that's something you can instrument, but it likely requires a lot of channels from a lot of different tiles.

(Also if you did have a black box, you'd just transmit the data after comm blackout and before it impacts the ground. No parachute, no landing, simple batteries instead of solar panels, high power omni transmitting to nearby orbiters, much much much simpler than a lander)

That's a mighty fancy antenna you're envisioning.  Remember, your black box needs to not burn up, so it's not exactly the ideal surface.  Plus, what's the bandwidth ground-to-LMO for the existing birds?  Enough to get enough channels of stuff spooled up before the box hits the ground?

[sdsds]

I'm confused. This ship can deliver 100t of payload to LEO, unless that payload happens to be prop, in which case it can deliver only 87t?

Thanks to you both for those clear explanations of the error in my thinking. If I'm understanding correctly a 'vanilla' Starship with zero payload arrives in LEO with > ~85t of prop. If two tankers could then each deliver another ~85t of prop and assuming no boil-off the ship would then have ~255t of prop. From LEO to a Vinf of 4 km/s requires ~250t of prop? That gets them to Mars with a departure on the last day of 2026.

[Twark_Main]

"Hypersonic reentry telemetry" is not a feature of "vanilla interplanetary comms."

Hey, you're the one who wanted to insult my hobby horse.  I was only talking about the vanilla stuff.

"Hypersonic reentry telemetry" is a quote directly from your earliest post, so you were trying to insert the hobby horse's nose into the tent from the very beginning. Don't get mad at me you got caught.

Tech we mostly already have is using existing orbiters as relays, and existing imaging assets to monitor the reentry.

That's fine if you happen to have an orbiter that you can send to through the plasma sheath.

It's also fine if you don't. Plenty of Mars missions have dealt with the Seven Minutes of Terror, and they didn't cancel or delay missions over it. Starship is no different.

Again, this is pure hobby horse self-insertion. Wants != Needs.

To paraphrase a famous comedian, "how quickly you feel entitled to something that [spaceflight] only knew existed five minutes ago."

Even just knowing the exact time of breakup and the probable breakup mode (how many pieces) would be a hugely valuable data point, and way more information than most Mars landers ever had access to.

No, it's exactly the same info they generate via the tones. One can quibble about how many tones you have, but JPL is pretty confident in their current lander architecture.  Starship, not so much.

The big unknown for Mars is how the TPS is gonna oxidize.  I imagine that's something you can instrument, but it likely requires a lot of channels from a lot of different tiles.

"But... like... I really really want it!!" is still not a Need.

(Also if you did have a black box, you'd just transmit the data after comm blackout and before it impacts the ground. No parachute, no landing, simple batteries instead of solar panels, high power omni transmitting to nearby orbiters, much much much simpler than a lander)

That's a mighty fancy antenna you're envisioning.

Fancy? No. It's just an omnidirectional antenna, literally one of the most basic kind.

Powerful? Yes.

This is one of those "upmass cures all ills" situations.

Remember, your black box needs to not burn up, so it's not exactly the ideal surface.

Yeah, I was picturing just a small CF sphere covered in PICA. A mid-sized intern project for SpaceX.

Plus, what's the bandwidth ground-to-LMO for the existing birds?  Enough to get enough channels of stuff spooled up before the box hits the ground?

Bandwidth is a function of transmit power. 

"Enough?" Sure, because naturally that would be precisely the design goal. And fortunately, failure is an option in this case.

[Vultur]

You are just repeating discussion several pages ago, where I said chance of Starship being launched towards Mars in 2026 is 0%.

Yup you said that.

At any point in the last 20 years, there were those who were betting it'll never work, it'll be decades away, it won't be viable.  (For various values of "it")

And the caravan kept moving.

So, we'll see.  I can't estimate the probability of a smooth v3 up-ramp.  But I am pretty certain that if v3 flights go smoothly, we will see a 2026 Mars intercepting flight with an EDL attempt.

I'm also pretty sure that if v3 ships continue to RUD, then they won't.

Well no you definitely won't.
It will require about 5 tanker refills.
A lunar surface and takeoff demonstration in 2026 will require at least that many.
Plus Starlink flights, and not to mention actually getting to orbit for the first time.
It is obvious that in designing the worlds largest ever launcher and making it fully reusable, there is going to be a considerable difference between the projected timeline and reality. As has been demonstrated many times.

I don't think they'll let anything get in the way of the window.  They might make some flights dual-purpose, but otherwise they'll push for Mars.

Starlink can continue at 250 v2.mini on F9 just fine for another year.

Regarding moon, all they promised is that they won't be the long pole.  I don't think there will be a shortage of long poles...

Basically - don't count on distractions on derailing the campaign.

Your 5 refueling flights - that's after the 8-9 months of development.  That's the launch campaign proper, starting September-October, and utilizing two pads, if development was successful enough.

All they need is to get to reliable orbital flights (2-3 good flights) then simultaneously iterate on refueling and EDL..  this can fit in 8-9 months.

You are aware aren't you that Artemis 3 is way behind schedule and the unmanned demo to the lunar surface has to be successfully completed before the manned version, so 2026 demo is really the latest it can be.

Not sure.
Artemis I - Nov 2022
Artemis II - maybe April 2026? (3 years 5 months)

Artemis III could easily slip to 2029*, so I'm not sure a 2027 HLS demo is that unreasonable.

*(Will the spacesuits be ready? What problems will be discovered with Orion during Artemis II? How will the post Artemis I heat shield modifications work out?)

[SpaceLizard]

You are aware aren't you that Artemis 3 is way behind schedule and the unmanned demo to the lunar surface has to be successfully completed before the manned version, so 2026 demo is really the latest it can be.

Not sure.
Artemis I - Nov 2022
Artemis II - maybe April 2026? (3 years 5 months)

Artemis III could easily slip to 2029*, so I'm not sure a 2027 HLS demo is that unreasonable.

*(Will the spacesuits be ready? What problems will be discovered with Orion during Artemis II? How will the post Artemis I heat shield modifications work out?)

Speaking of Artemis's spacesuits, has there been any news about them lately?

[sdsds]

Speaking of Artemis's spacesuits, has there been any news about them lately?

None have been reported on this thread, for good reason! ;-)

FWIW my model says Starship to Mars will happen independently of Starship HLS demo for multiple reasons.

* Dis-similar vehicles (HLS demo needs landing gear, and likely wants surface cargo deployment)
* Different pads (my model says NASA wants HLS launching from the Cape)
* Different cultures (Mars is development heavy 'cowboy' risk, HLS is production quality caution)

[TheRadicalModerate]

"Hypersonic reentry telemetry" is a quote directly from your earliest post, so you were trying to insert the hobby horse's nose into the tent from the very beginning.

So I did.  I don't remember adding that. Apparently the hobby-horse will not be denied access to the fingers as they type.

I think the black box is worth exploring.  But if you're not gonna let it land, then it has to transmit all its goodies to one (or more) of the birds in the relay network before it hits the ground.

I'm gonna do a little research and post something over on the comms/PNT thread.  It's a question of what peak surface-to-LMO bitrates are supported by MRO, MAVEN, Odyssey, and the ExoMars TGO.  It's also a question of how reliable you can make the ejection mechanism.

That's fine if you happen to have an orbiter that you can send to through the plasma sheath.

It's also fine if you don't. Plenty of Mars missions have dealt with the Seven Minutes of Terror, and they didn't cancel or delay missions over it. Starship is no different.

Starship is different because it's a test article, not a science mission. SMOT-style entry vehicles are designed very conservatively, because the payload is the point of the mission.  That's why they're so wedded to low-ballistic-coefficient, blunt lifting bodies, with roll-based guidance.  It's also, not coincidentally, why everybody's freakin' out about the MSR SRL, which has to be bigger, with a higher BC.

But the SRL is child's play compared to getting Starship EDL to work.  If you're going to go the expense, both economic and political, of jamming a Mars mission into the middle of most crucial part of the HLS development cycle, you want to get the data you need to dramatically improve the odds of success in 2029.

That makes it seem a lot more like "needs" than "wants".  Certainly getting enough engineering telemetry is a need, not a want.  However, maybe 128Kbps is a need, and 100Mbps is a want.

Maybe Elon is OK with a PR stunt.  But if that's the case, it's a much better PR stunt with HD video.  That is, of course, a lot more bits than engineering data.

I still think the Trailer is easier and better than the black box.  For the kind of lifting trajectory required to get a Starship to decelerate, the difference between the entry corridor and the hyperbolic flyby periapse is only going to be a couple of degrees.  You need the Trailer far enough behind that it doesn't overtake the Starship as it slows down.  How far behind would be an interesting calculation.

[TheRadicalModerate]

FWIW my model says Starship to Mars will happen independently of Starship HLS demo for multiple reasons.

* Dis-similar vehicles (HLS demo needs landing gear, and likely wants surface cargo deployment)
* Different pads (my model says NASA wants HLS launching from the Cape)
* Different cultures (Mars is development heavy 'cowboy' risk, HLS is production quality caution)

I think they're coupled because of tanker cadence.  Getting prop to the depot as quickly as possible reduces boiloff losses, so it probably makes sense to dedicate all pads to launching the necessary number of tankers as quickly as possible, then putting your payload ships (i.e., your Mars Starship and the HLS) on the pad as soon after the tanker campaign is complete.

That means that you can't really conduct two decoupled campaigns.  You do one, then you do the other, because they're sharing the same resources (tankers, pads, maybe depots).  You can't just interstitially jam a Mars mission into the middle of the Option A test or HLS campaigns.

Things are substantially different if depot boiloff losses are very low.  Then tankers can be launched interstitially between Mars, HLS test, HLS crewed, and Starlink missions.  In that case, I'm more likely to buy your "two independent efforts" model.

[sdsds]

FWIW my model says Starship to Mars will happen independently of Starship HLS demo [...]

I think they're coupled because of tanker cadence.

From a technical perspective you are correct. That said, NASA human spaceflight has not historically been driven by what's technically optimal.

Pretty clearly if HLS needs launches from Starbase they're going to get them. My prediction is that when we first see a 'plan of record' for HLS-related launches they will all be from the Cape, and there won't be more than fig-leaf cover for that from the technical side.

If HLS does need Starbase launches, the likelihood of Mars in 2026 goes down markedly.

[Vultur]

Even if Mars is totally written off, I'm actually still rather skeptical of HLS Demo in 2026. I think that probably has a lower tolerance for failure, and it requires actual landing and I think now takeoff too; probably harder than an interplanetary cruise test.

If v3 goes very very smoothly, it's possible, but... Honestly I think there's a feasible path where in Nov-Dec 2026 they feel ready for an interplanetary cruise test that is ok with a high probability of failure, but don't feel ready yet for HLS Demo.

OTOH, there's also the possibility that they do an earlier Moon pathfinder that isn't the official HLS Demo.

[TheRadicalModerate]

Even if Mars is totally written off, I'm actually still rather skeptical of HLS Demo in 2026. I think that probably has a lower tolerance for failure, and it requires actual landing and I think now takeoff too; probably harder than an interplanetary cruise test.

I'm also skeptical that the Option A demo will occur in 2026.  But it needs something like 6 tankers, even for the minimal ascent test.  A lot obviously depends on tanker launch cadence, but if they plan to do the demo in 1Q27, there's a pretty good chance that they need to be devoting those tanker resources in 4Q26, maybe even 3Q26.  They can't do that if they're busy trying to make a November-December Mars launch window, complete with 3-4 tankers.

Just to get a feel for how much they need to improve:  In 2025, cadence currently averages 1 launch every 74 days.  At that rate, in order to hit a demo by the end of 1Q27, they'd have to start launching operational tankers in early January, 5 months from now.  And that assumes zero boiloff.