SpaceX wins first Pentagon contract for Starshield | Hacker News Reader

SpaceX wins first Pentagon contract for Starshield

cnbc.com

54 pointsperihelions64 comments

That this will happen was obvious in March 2022, the only unknown was price. Starlink has proven itself to be unjammable, resilient, low-latency, high-bandwidth, reliable, easy to setup and all that confirmed in a real kinetic war.

Basically everything any Pentagon official could have dreamed of, for $70M? Probably the fastest check ever written there (yes yes it's just the beginning but you get the idea.)

Umjammable is not guaranteed. It’s a new tech (which is/was also geofenced from parts of the front line), so we there wasn’t that much time yet for adversaries to develop jamming tech.

It’s very likely harder to jam, but unjammable is a very bold statement.

Jamming ultimately works by known physics though, and Starlink seems like inherently an incredibly tough thing to jam (even ignoring anything in the future like optical). It's very high frequency, in LEO, using phase arrays and electronic steering. That makes for something that's highly directional, fairly tight beams, fairly strong signal, that are heading up into the sky (in a military context meaning anything trying to get between you and the sats is right above your SAMs/AA/fighter screens), moving very fast, with the built-in (needed anyway) ability to near instantly jump between satellites in different directions of the sky. That's a big list of fundamental physical challenges that omnidirectional low frequency radio or single HEO sats or the like just don't have. It will likely not even be reasonable for a long while to shoot down the constellation (retaliation aside). SpaceX's launch and mass manufacture costs are already cheaper than anyone and Starship will take that down another order of magnitude or more. Starlink sats will be cheaper than ASAT missiles, and orbiting low enough that atmospheric drag means Kessler Syndrome is not a concern either. Attacking the launch site(s) means attacking the continental USA.

It seems to add up to a pretty dang hardened looking system? Cyberattacks, sabotage and so on could be real concerns, but to jam a tight beam phased array looking at very specific spots for signal from 10-100 miles away looks hard from a physics point of view. And of course any powerful jammer will by definition be an absolute magnet for HARMs. The US takes SEAD/DEAD very seriously and is thus pretty heavily equipped for that.

> Jamming ultimately works by known physics though, and Starlink seems like inherently an incredibly tough thing to jam (even ignoring anything in the future like optical).

Jamming DownLink is certainly difficult especially if ground terminal uses receiver beam steering and pointing upwards. Jamming UpLink is easier by beaming directly at the satellites' receiver antennas. Jammers can use beam steering too.

Even ignoring a number of other tricks it could be using:

>Jammers can use beam steering too.

Beam steering doesn't make the beams curve though. The satellites know where the terminals are, and last I checked they do 8 or 16 beams and current beam spot size is 22km in diameter (and set to shrink with full size v2 and future, that's key to improving density of service). Which means your jammer would need to be tracking the right satellites without help from the satellites themselves (unlike real terminals), and within 22km (or far, far less, and on the correct side of the cell) of the terminal. I invite you to compare that to the range of basic 155mm artillery, GMLRS missiles, JDAMs, HARMs, or other such weapon systems, and remember we're talking about this in a military context. Is a jammer that has to be so close it gets destroyed right away militarily significant?

I agree. A more precise statement would be - as of now, Russia does not have the tech to jam it. They must be working on it and same for the Chinese, but for now, we haven't had any evidence that jamming is operationally deployed.

Do you mind to clarify what do you mean with "real kinetic war"?

Whatever is going on in Ukraine for the past year and change. As opposed to trade wars or cyber warfare.

Its a fancy term for a hot war in meatspace.

Ukraine, this isn't a cold war — but a "real war" with munitions (kinetics) exchanged

A war where the bullets fly.

So they’re just creating a vlan named “starshield” and then depositing $70M in the checking account?

Must be nice.

;)

IIRC one big feature of Starshield (compared to the source of much of the drama and false reporting regarding Starlink in Ukraine) is that the buyer (ie the DoD) controls the cells it's active in.

> So they’re just creating a vlan named “starshield” and then depositing $70M in the checking account?

It's always difficult to price products and services with high fixed costs and low marginal costs (like bridges). The best estimates I've seen are that Starlink costs $1B+ to operate. It sounds like they're finally profitable (at least on an ongoing basis); we'll see what happens when Kuiper finally comes online.

I'd say Kuiper is still firmly an "if" right now considering how long they have left to put up the constellation and the cost+potential flight rate of Vulcan and New Glenn. They have a much higher bar to clear in terms of costs before they can be profitable, and especially with the lawsuit about that, I wouldn't be surprised if eventually they decide to cut their losses and settle for a smaller constellation and restricted user base like OneWeb.

> I'd say Kuiper is still firmly an "if" right now considering how long they have left to put up the constellation and the cost+potential flight rate of Vulcan and New Glenn.

I think there is pretty broad consensus that they won't make their deadline. But I also see pretty broad consensus that, if they make a strong effort, that a variance is pretty easy to get. I'm not really in the industry, so I don't really know what to believe.

> They have a much higher bar to clear in terms of costs before they can be profitable, and especially with the lawsuit about that, I wouldn't be surprised if eventually they decide to cut their losses and settle for a smaller constellation and restricted user base like OneWeb.

I always saw Kuiper as a strategic move by Amazon to attack SpaceX. IMO, they'd probably scrap the whole thing rather than settle for a OneWeb sized constellation.

More broadly, I think satellite internet is a natural monopoly business. There's no reason for customers to go with the 2nd best service[1], so even though it costs more to launch the best constellation, it ends up being a better investment. Because of this, I don't really see OneWeb or Telesat's upcoming Lightning LEO constellation surviving longterm. There's just no good reason for their existing customers to be loyal to them. And the cost to maintain their constellations will be ongoing.

---

1. Obviously globally available bandwidth affects this, so there's probably room in the market right now for multiple providers. But as companies continue to invest revenue into improving their capacity, those conditions may go away. At which point, my original point stands.

While overall I agree with your points, the reason I think they might settle with a smaller constellation is that they could sell it as an AWS service with direct integration with their data centers. For example, as a lower latency low bandwidth interlink between continents. It would be similar to how Starlink integrates with the networking at Google datacenters, but made explicit as an AWS service.

Price would be definitely something that could persuade me to take the number 2 service

> Price would be definitely something that could persuade me to take the number 2 service

Sure. I guess what I was trying to say and didn't really do a good job of it is that the best network, if they price their service properly, should undercut everyone else. That's basically what it means to be the best, that they can offer global bandwidth for less money than anyone else.

The announcement article linked within discusses imaging and other hosted payloads.

At only $70M it's doubtful this specific contract includes any dedicated satellites. A single Falcon 9 launch is close to that price.

> At only $70M it's doubtful this specific contract includes any dedicated satellites. A single Falcon 9 launch is close to that price.

While I broadly agree with you, if SpaceX is selling a service and not a launch, I think the proper number is the internal cost to launch which probably sits around $20m-$25m.

You are about right on cost. Elon stated that a full reuse of a first stage + new second stage is about $15m, adding in the Elon-number correction and 20-25 feels right.

IIRC the $15m is the per-unit costs (booster refurb, second stage production), the rest are the costs that can be divided up by increasing flight rate (launch pad operations, staff and other infrastructure).

SpaceX has a very large fixed cost base, wages and facilities mostly, and an infinitesimally small variable cost for a rocket launch. Things like RP-1 fuel, catering, aluminum coil, actuators, valves, boards, wiring, connectors, airfares, hotels, taxis, etc. The upper stage construction line and launch facilities are fully amortized by now and from the outside seem like highly efficient operations.

Given that their launch cadence has doubled since people started throwing around $25m estimates and fairing and first stage inspection and reuse is now routine it's probably a bit lower.

Why would this need a dedicated launch? Starlink satellites launch in a stack of up to sixty at a time.

Starshield probably just swaps one or two of these for an alternate configuration (i.e., same bus, different payload).

I think creating then launching thousands of satellites cost-effectively is pretty hard

Interesting its per year. Elon did a tweet a few weeks ago that said it would be "owned" and "controlled" by the DOD.

https://twitter.com/elonmusk/status/1707149214701019392

Owned and controlled probably just refers to the DoD being allowed to do practically anything it wants with it and having the controls to enable/disable cells. So no restrictions like "don't mount the antennas to drones" and no need to request SpaceX to enable/disable cells.

I'll be somewhat curious to see exactly what this entails, at least to the extent we ever get any details. One aspect could be another facet of Starlink that I think still hasn't gotten enough appreciation: the possibilities opened up by a heterozygous orbital mesh network. The orbital mesh of course is primarily first order about making service available from LEO worldwide, including areas too far from any possible ground station to allow a simple bent pipe. Second order is more redundancy and load balancing, lower latency for long enough distances vs terrestrial networks, etc, which has all gotten discussed. But the physical nature of it means that once tech and protocols are there for the laser mesh, that acts as a standardized interface that any satellite could "plug into". They're of course needed to do have this in mind from the start due to how fast they've been iterating, with laser v0.9/1.5/mini-2.0 and full 2.0 once Starship is going sats all needing to talk to each other. But that means SpaceX could also easily mix in some percentage of Starlink sats that do other stuff entirely. They could have optical or radio bridge sats facing "up" for example, so that other higher satellites, space stations, probes and so on can talk directly into Starlink from medium/high Earth orbit or beyond. That could be an alternative to the Deep Space Network that could be quite attractive particularly as the DSN has been getting seriously overloaded [0]. Which also means it could be an important thing for SpaceX in driving cadence for Starship, all the stuff they want to launch and open up with it will need a lot more bandwidth back to Earth.

But they could also potentially have dedicated ground facing sats for a customer like the US DOD. Just spitballing, but perhaps ones that use broadcast frequencies forbidden to commercial providers. Or even ground/space optical (at least for aircraft, clouds would be an issue below), which would offer tens of gigabits to terabits of bandwidth and be unjammable. That would offer a range/reliability boost in a remote theater for the 1st Earth <> LEO leg of things, then plug right into the overall mesh for comms back to the US via the normal sats, so it'd be cheaper, lower latency and higher bandwidth then military trying to do their own completely physically separate thing. Distinct capability for less.

----

0: "NASA officials sound alarm over future of the Deep Space Network" https://arstechnica.com/space/2023/08/nasas-artemis-i-missio...

One step closer to Reagan's Strategic Defense Initiative: https://news.ycombinator.com/item?id=32839690

Anything preventing Musk from building these for other countries?

The fact that he's domiciled (and seems to enjoy his life) in the US, and that the government has the ability to take everything you own and lock you up if they so decide. Probably.

That works so long as the government still has more power than a gang of bajillionaires with similar interests. Does it? Seems like we're near a tipping point in that regard, if we haven't already crossed it.

> Seems like we're near a tipping point in that regard, if we haven't already crossed it.

Sure... As long as the army is following the government it's a non-issue. Billionaires are political/industrial/&c. power if you want but they're one "go get them" order away from disappearing forever. It just so happens that for now their goals are somewhat aligned with the gov's.

And even if the gov falls a lot of US three letters agencies are already semi or fully autonomous, and from the few things we know about them they don't fuck around.

Corruption exists, probably on a large scale, but it's not a single coordinated effort, it's more like an army of rats fighting for the crumbs. If even Mexico didn't completely fold against their drug cartels it's not the soy boy Musk & lizard man Zuck dream team who'll destroy the USA

I think US defense blob is actually the entity at the top of the entire global power food chain. Probably the actual last entity you'd want to meaningfully fuck with.

As Mao once noted, at the end of the day, political power grows from the barrel of a gun. Billionaires certainly have significant influence over policy decisions, but the tools of violence are still controlled by the politicians. Money can't protect you from the army or the secret police, as Jack Ma discovered in China.

Even were it the case that the US military could be used against a US citizen, your observation would still be cold comfort in our world where billionaires can buy politicians. Worse, the Citizens United SCOTUS decision equates such control with speech, and is tacitly protected by the government.

Billionaires aren't a monolithic block though (beyond core issues like low taxes). If you sell out your country, then other oligarchs can use that as an excuse to order their pet politicians to confiscate your fiefs^H assets. Just see all the speculation about nationalizing SpaceX in this discussion.

The only way money can protect you is if you literally buy out the entire government, in which case you're basically a monarch.

But yes, that is cold comfort to us plebians.

Elon Musk has firing control of 16 rockets that are trivially convertible to ICBMs. It's not a standing army, but it's one helluva gun.

Yes, it does. The government, indirectly or directly can indefinitely owe any entity money, or even print money on demand if really necessary and ultimately, Musk is just a person.

The $5 wrench can be a pretty big motivator

Defense technology export restrictions would get in the way, so at most I guess they might be able to sell this to particularly close allies like the UK and Japan.

ITAR [1]

Not that the Chinese are perfectly capable of building the same network on their own. Albeit with electronics' performance a transistor node older. Which is to say: almost the same.

[1] https://en.wikipedia.org/wiki/International_Traffic_in_Arms_...

The same thing preventing Lockheed from building F-35's for other countries: national security interests.

I'm sure they'll be allowed to build commercial satellites for other countries that contain a kill switch the military has access too. Military hardened satellite networks will be a non-starter other than maybe for some close allies like the UK.

Maybe. It would depend on how the government classifies it for export controls. The fact they've signed a contract leads me to believe that satellite communications in use by the government will be an ITAR control.

https://en.wikipedia.org/wiki/International_Traffic_in_Arms_...

I suspect the US gov will just take SpaceX off him if he does anything objectionable.

Like own a social network that could allow anti-mao speech but has taken covert steps not-to and have a Chinese car factory that makes up a large percent of his auto profits? I'm kinda shocked that they allow such a leveraged person have contracts like this where other countries have levers they can turn.

He's a useful idiot at the moment. Once he is no longer useful, he's collected a very long rope.

I suspect the end game will be Tesla's collapse when his hype train hits a wall and other manufacturers perfect low cost vehicles, Twitter/X is already dying and SpaceX will get involved in an ITAR violation or conflict of interest and suddenly end up in the hands of Lockheed Martin or someone similar and him in front of the SEC for one last time.

The US gov would not want to look like they're intervening. They don't need to.

I fully expect this will happen.

USG has more control if it's done by Starlink vs some other country's tech.

A US federal employee with a $5 wrench.

ITAR

DOD must have plenty of assurance that Musk isn't a significant part of the equation. The dude brings so much risk with him.

Maybe this is the way they can nationalize spacex once Musk does his next stupid thing

This is such a weird obsession you guys have.

Notice how the US government has never commented on any of his "stupid things" (with most of the Ukraine related ones turning out to be either lies or misleading reporting). Hell, US policy on that front has until extremely recently with the supply of ATACMS been fully in line with the restrictions placed on Starlink by SpaceX.

The government is not going to nationalize SpaceX for such basic reasons. Defense contractors and other large companies regularly go against the government with much larger sums of money involved without any of this delusional thinking getting involved.

Reply on news.ycombinator.com