Investors commit quarter-billion dollars to startup designing “Giga” satellites

"If we build these platforms well, we get to ask new questions about what's possible in orbit."

See full article...

 

[rochefort]

If launch costs are $5M for 100 tons, or $50 / kg, then this is $20,000 of launch cost per chip.

Which seems unlikely anytime soon.

 

[Ceedave]

Space itself has no temperature. So what you’re relying on is radiation to “cold areas” and that goes as T^4. So raising your radiator from 300K to 400K makes a hefty difference (about x3) in the area required. If you have to radiate kilowatts or megawatts, spending some energy and mass to “pump the heat uphill” to a higher temperature might be worthwhile. Right now, traveling wave tubes often use radiation cooling - they’re vacuum tubes and running hot is just fine, so they don’t need a big radiator. Similarly, one of the positive things about Gallium Nitride semiconductors is that they are perfectly happy running at 130C, so radiating the heat requires 1/3 the mass of radiator.

OTOH, if “mass is cheap” then the trade is less clear - but you DO get into a scaling problem. Enlarging a radiator area by a factor of 10 makes all the structures 3x longer which means they need to be heavier to support the loads, and then there’s the conduction of the heat issue.

Space has very poor capacity to transport heat, but various components and regions of interplanetary space do have well defined and diverse temperatures, as explained in this well-sourced Wikipedia article. Because the sparse prticles in the imperfect vacuum of interplanetary space have different modes and/or speeds of motion, temperature varies as well. Dust particles in near Earh orbit are about the temperature of a nice fall day in New England, and cislunar solar wind is about 10,000 K. Every speck of matter has a temperature, and space has matter in it, just less than we are used to.

 

[Klive Aleksaander]

(I don't post often enough to understand why I can't add a link to Wikipedia here, but it's the first hit)

You may need to have posted a certain number of comments here before you can add links.

Edit: And I see you have only made five comments so far, including the one I have responded to.

 

[kaleberg]

So aside from being big and having lots of power, what do these satellites do?

Jewish space lasers and just in time for Hanukah.

 

[SportivoA]

Apparently the current recordholder for longest duration operational satellite is AMSAT-OSCAR 7. Launched over 50 years ago and still sending signals as of the last Wikipedia update in March.

It's a really cool bird, but I'd say it had a pretty big impairment on the "continuous" operation side. One day I'll actually be set to work it.

Edit: And I see you have only made five comments so far, including the one I have responded to.

And I think that's one limit to start adding links going forward. Though there are plenty of other spam techniques to attempt the craft on these forums.

 

[DDopson]

Space has very poor capacity to transport heat, but various components and regions of interplanetary space do have well defined and diverse temperatures, as explained in this well-sourced Wikipedia article. Because the sparse prticles in the imperfect vacuum of interplanetary space have different modes no speeds of motion, temperature varies as well. Dust particles in near Earth orbit are about the temperature of a nice fall day in New England, and cislunar solar wind is about 10,000 K. Every speck of matter has a temperature, and space has matter in it, just less than we are used to.

Sort of. As the wiki says...

Because the interplanetary medium is so rarefied, it does not exhibit thermodynamic equilibrium.

Each dust spec has enough particles in close contact to form a local thermal equilibrium, which more formally means that if we make a histogram of the relative particle velocities within a dust spec, they'll conform to the Maxwell Bolzman distribution for some temperature. But then, the dust specs aren't in equilibrium as a collection. Their temperatures vary, and their orbital velocities around the sun are very large compared to their local thermal velocities, and this becomes obvious whenever two dust specs on materially different orbits manage to collide, which can easily heat their combined mass into a plasma.

Things are even weirder for the individual particles wizzing around because they are more like the collection of rarely colliding dust specs than they are like the local thermal equilibrium established within a dust spec. Also, a free-flying particle neither has a defined temperature, in the absence of interactions, nor can it emit thermal radiation without having other particles to interact with, which means it can't "cool down". Temperature is a statical relationship between particles, not a property that an individual particle can possess.

For example, in the upper part of Earth's thermosphere, around 600 km, there are particles moving so fast that you can find pages quoting temperatures of 2000 C, but compared to a M-B distribution, the velocity distribution of the particles in the upper thermosphere has a very long tail, with more ultra-fast outliers than would be expected for a properly thermalized population. What's happening is that EUV photons from the sun rip molecules apart and send individual atoms and free electrons flying off at high velocities, restocking the population of particles with velocities characteristic of absurdly hot temperatures. Those particles can go entire seconds between collisions, so it takes them a long time to thermalize, and apparently the actual collisions do less to radiate away energy than the vibrations induced in diatomic molecules like O2 and N2, which have the ability to self-interact in between collisions with other molecules / particles. And then the Earth's magnetic field causes ions to move faster in some directions than others, adding yet more complexity not captured by a vanilla thermal description. I don't pretend to exhaustively understand all the various bits of physics that are in play, but the point is that temperature is insufficient to characterize the motion within that collection of particles.

The other thing that's very far from equilibrium in space is the photon flux. The black parts of space are 2.7 K, but the overall flux is utterly dominated by the 0.15% of the sky covered by the sun at 5700 K. The other stars are pretty hot too, but the total angular area for all other stars combined is much less than that for the sun, so the solar radiation dominates. The Earth is a bit funny because it's radiating with a thermal distribution of ~300 K, but the sunlit side is also reflecting 1/3rd of incident sunlight, including a lot of visible photons that are way more energetic than anything coming from 300 K thermal radiation. You can confirm this by observing that we can see the day side of the Earth from space, in the visible wavelengths, but then the night side of the planet is pretty dark because even the warmest parts of the Earth are nowhere near hot enough to glow in red, much less orange or yellow (city lights, lava, and lightning excepted).

Larger blobs of matter in space at 1 AU will generally be pretty cool because their internal interactions enable them to radiate heat away until they reach an equilibrium with the average photon flux they are being exposed to, but the extreme non-uniformity of the photon flux means that the equilibrium temperature is incredibly sensitive to the object's shape, orientation, and relative albedo at various frequencies. For a spherical black body at 1 AU, but distant from Earth, the equilibrium temperature is 5 C. Or in a dawn/dusk orbit it's about 11 C. On the sunny side of a LEO orbit (500 km), 33 C, on the night side, -84 C, or with enough thermal mass to average across the orbit, -19 C.

Or if you paint a sphere with one of those special coatings that's black in the IR frequencies and white in the visible and higher frequencies that dominate solar radiation, much lower temps are possible, -82 C in deep space at 1 AU, or -47 C in the dawn/dusk orbit (the Earth's IR flux becomes much more relatively significant when the object is reflecting most of the solar radiation).

Or if you make your object long and thin, and orient the thin part towards the sun, you can much closer to 3 K (I think the average with starlight is ~3.2 K, and you also can't avoid all sunlight because the sun has a non-zero width).

 

[DDopson]

Which seems unlikely anytime soon.

Sure, $5M for 100 tons is an aspirational cost for a Starship-like launch platform that's operating continuously at scale. Each flight is about $3 million in fueling costs, so to stay under $5M in total costs requires extreme amortization of the launch vehicles, launch site, and operational labor, but it's not a wildly unrealistic target if the flight cadence is very high. Not that they'd sell a launch for that price, but that they could plausibly achieve that internal cost from a continuously operating spaceport that launches many times per day.

Anyways, the whole point was to use a fairly rosy number for launch costs, because even with rosy numbers for launch costs, it still takes about 40 years for the electricity savings to pay back the launch costs. On datacenter hardware that will need to get upgraded every 3 to 5 years.

It's much, much easier to get to $5M Starship launches than to design an orbital datacenter that's cheaper to launch than the solar+batteries required to power it with the sunlight that reaches Earth.

 

[NerdInMA]

Thanks! And just personal for now. Haven't really found any place to post short stories outside of Reddit these days (and I'll never go back to Reddit), and no one really trusts personal sites any longer.

Most of the small/new writers I follow have personal sites - I am not seeing any trust issues anyone would have going to a personal site, but maybe I am missing something. Most of these writers (but not the same most in the previous sentence) also use Patreon, and distribute that way. Your writing seems quite enjoyable with what I have seen so far, and I would enjoy reading a short story or longer from you.

 

[thesandbenders]

provides a standard platform for others to build on. So the others don't need to spend time and effort creating their own version of the basics that this provides. The others can specialize in creating the compute or telescope, etc.

Got it one but a little more detail. The industry term is a bus and the top 5 buses have been used in at least 383 satellites.

 

[Cart_catalog]

Maybe I'm just having a conditioned response to giga this and mega that. Bullshit buzzwords for product names always make me wonder about the quality of the snake oil that's being sold.

AMEN! Enough with the buzzwords already. Maybe think up your own terminology for a change?

Soon 'mega' and 'giga' will be meaningless words that tell us nothing, if they are not already.

 

[ingramm]

More targets for the Kessler syndrome ......

 

[Astra Architect]

I'm just a simple engineer. A person of the land. The common clay of New Space. You know...a moron.

Let's look at the typical use cases for satellites: communications (TV/video, Internet, telephone, radio); positioning, navigation, and timing (PNT); Earth observation (EO); military purposes (usually the first three, just government-only use); emergency services (often a combination of the first three); and science observation and research.

These mega-satellites only make sense for non-bespoke science missions if K2 is just the bus provider and management expects companies to buy one or more payload slots on the bus.

Absent some pretty significant investment in segregating circuits/data flows, and getting veto power on who's allowed to ride-share on the bus, the U.S. government is not going to be keen on ride-sharing. DoD/IC might buy a bus and put only USG payloads on, but they will be very reluctant to ride-share classified payloads with commercial payloads. And by "very reluctant" I mean they won't do it unless the Orange Menace orders them to.

Also, the U.S. DoD and IC are deliberately moving away from large buses towards distributed architectures - harder to put a serious hurt on our capabilities if we have 50/100/etc. satellites for doing something.

If I were a commercial company then I'd be hesitant about putting my payloads on a ride-share with one or more USG payloads. We've entered the era of overt space warfare. We demonstrated space warfare capabilities decades ago, but didn't really talk openly about space as a core warfare domain for until the last few years - by which I mean as a domain where we discussing taking direct actions, not a domain with operations only to enable actions in the other domains. We get in a shooting match with Russia or China, or even a "gray zone" conflict, and those satellites are targets. I'd sure hate to lose $20M/$50M/$100M+ on the hardware and whatever the cost in lost profits is if my satellite bus is targeted because I ride-shared with some USG payload.

And who do I even go to for my losses? Insurance policies typically exclude coverage for losses or damages due to "acts of war" when it comes to property, business interruption, and general liability. Acts of terrorism are usually covered. But any insurance company with a modicum of business sense will fight having to cover a loss due to a satellite incident as an act of terrorism - the odds of it coming from a terrorist group that was not sponsored and authorized by a nation state are vanishingly small, and the company could pay a lawyer full-time for years before the legal bills exceed the value of the claim.

Do I sue the other nation-state? That's a waste of my time. I'll be lucky to see a penny. To my knowledge there are only two occasions of nation-states agreeing to pay legal settlements for terrorism against U.S. citizens: Libya (PanAm Flight 103) and Sudan (embassy bombings in Kenya and Tanzania, USS Cole, and a murder of a U.S. development worker in Hartoum). Billions of dollars have been awarded in judgments against Iran, North Korea, Cuba, and Syria without a single payment being collected. The two countries that settled did so out of desperation for aid and wanting to attract foreign investment.

Multiple commercial providers ride-sharing makes sense. There's no particular technical reason you couldn't do comms, EO, PNT, and even some science missions on the same K2 bus if you do proper EMI/EMC. Whether the orbital altitudes and inclinations make sense for a bird to have payloads supporting multiple missions is a different question.

 

[Stuart Frasier]

These mega-satellites only make sense for non-bespoke science missions if K2 is just the bus provider and management expects companies to buy one or more payload slots on the bus.

Nothing in the article indicates that they intend to make a small number of satellite buses for ride-shares. They are planning to sell to a supposed future market that is less mass-constrained.

“The idea is that K2 will manufacture the satellite chassis, and customers will use it to accommodate their own unique payloads. Examples of missions Giga can support include AI computing, high-throughput networks, and mass-produced giant telescopes for astronomy.”

Whether that market actually emerges is a different question.

 

[Veritas super omens]

The way things are going with the commercialization of space, I suspect that these ( and other satellites in graveyard orbits ) will eventually be retrieved and either recycled on-orbit or deorbited.

One thousand years ago steel blades were the latest and greatest human technology. Most humans couldn't read or write and vast swaths of the globe were the unexplored lands of dragons and sea monsters.

One thousand years from now we'll either be extinct, or quite capable of policing up space junk.

We still have sea monsters. But the spelling has changed. They are now C-suite monsters...

 

[Veritas super omens]

The way things are going with the commercialization of space, I suspect that these ( and other satellites in graveyard orbits ) will eventually be retrieved and either recycled on-orbit or deorbited.

One thousand years ago steel blades were the latest and greatest human technology. Most humans couldn't read or write and vast swaths of the globe were the unexplored lands of dragons and sea monsters.

One thousand years from now we'll either be extinct, or quite capable of policing up space junk.

Steel blades were cutting edge technology? (Hah) actually quite a bit further back than that IIRC.

 

[Z06 Vette]

So a Falcon 9 will put a sat into orbit called Gravitas and then land on a barge called A Shortfall of Gravitas... sounds about right

 

[Astra Architect]

Nothing in the article indicates that they intend to make a small number of satellite buses for ride-shares. They are planning to sell to a supposed future market that is less mass-constrained.

“The idea is that K2 will manufacture the satellite chassis, and customers will use it to accommodate their own unique payloads. Examples of missions Giga can support include AI computing, high-throughput networks, and mass-produced giant telescopes for astronomy.”

Whether that market actually emerges is a different question.

Investment only makes sense if ride-shares are explicitly intended as a market alongside of whatever blue-ocean market they think they can create. It's like the 747.

Boeing's CEO started the concept phase on a handshake agreement with PanAm's CEO, IIRC...but Boeing's board told the CEO they'd only agree to fund the design phase if the design included accommodations to make a cargo variant as well as a passenger variant. The board knew there would be a demand for a 747 freighter, though how many 747s freighters they didn't know, so they said "design for both so we can sell more planes and get enough ROI to justify the cost - we don't think there will be enough of a demand for 747 passenger aircraft." The top deck of the 747 only exists because it was there to accommodate the machinery for the clam-shell nose door design.

Turns out Boeing's board was wildly off target on their estimates for the demand of both passenger and cargo 747 variants. But there was literally no data at the time to give anyone a clear enough crystal ball to know that.

Absent clairvoyance, nobody knows of a single viable business case for on-orbit AI that needs its own dedicated satellite bus and will be viable in the next 5-7 years. There's no business case for on-orbit data centers at this time. The heat management and radiation shielding requirements alone mean you'll end up with massive SVs relative to the computing power they provide. You'll either need massive shielding or you'll need lots of spare computing equipment that's not used until other equipment craps out.

Unless K2's investors are idiots - either ignorant of the aerospace market or outright fools - there's no way they're not telling K2 management to also try to sell buses for ride shares in the existing markets.

 

[Stuart Frasier]

Unless K2's investors are idiots - either ignorant of the aerospace market or outright fools - there's no way they're not telling K2 management to also try to sell buses for ride shares in the existing markets.

SpaceX only gets enough demand for 3 rideshare mission to SSO and 2 to LEO every year on Falcon 9. This giant bus on a heavy rocket would probably satisfy that demand with 2 missions annually and with plenty of space to spare. It's just not much of a market to pursue.

This company is hoping that the combination of a large spacecraft and cheap launch will inspire some sort of large constellations to justify the investment. Which may or may not happen. However, the small rocket market has been a bust. Only Electron is getting a respectable number of missions needing custom orbits or launch timing. Otherwise, the demand has been largely satisfied by a small number of rideshares on medium rockets.

 

[uhuznaa]

This is... Wrong, sorry.

Cooling things in space is surprisingly difficult because of how heat transfer works.

On Earth, we're used to cooling things through three methods: conduction, convection, and radiation. In space, you only have one option: radiation.

On Earth radiation is the only option too in the end. Cooling by conduction and convection just seems easier because you treat the atmosphere as an infinite buffer to simply dump your heat into. In the end it has to radiate all that to space anyway and you're really just using it as a transfer medium for that.

This is somewhat comparable to treating fossile fuels as an infinite resource and treating the atmosphere as an infinite buffer to dump your CO2 to. And we are already running into the limits of that after just one century of doing this at scale. Earth may seem big, but in fact it's quite small.

At scale there's a whole lot more solar energy to reap in space and there's a whole quite cold universe to radiate your heat losses to without ruining our atmosphere and bodies of water along the way.

Yes, this sounds like a crazy scale but people would have said the same about us being able to raise the CO2 levels in our atmosphere by just too much within an actually quite short time.

As soon as you don't just think short term and small scale all of this doesn't really look all that crazy anymore. In the contrary it will be the only option that supports scaling up by a lot in the long run.

 

[DDopson]

On Earth radiation is the only option too in the end. Cooling by conduction and convection just seems easier because you treat the atmosphere as an infinite buffer to simply dump your heat into. In the end it has to radiate all that to space anyway and you're really just using it as a transfer medium for that.

This is somewhat comparable to treating fossile fuels as an infinite resource and treating the atmosphere as an infinite buffer to dump your CO2 to. And we are already running into the limits of that after just one century of doing this at scale. Earth may seem big, but in fact it's quite small.

At scale there's a whole lot more solar energy to reap in space and there's a whole quite cold universe to radiate your heat losses to without ruining our atmosphere and bodies of water along the way.

Yes, this sounds like a crazy scale but people would have said the same about us being able to raise the CO2 levels in our atmosphere by just too much within an actually quite short time.

As soon as you don't just think short term and small scale all of this doesn't really look all that crazy anymore. In the contrary it will be the only option that supports scaling up by a lot in the long run.

Sure, but why limit your thinking to just one solar system? There’s far more energy available around the bigger stars that put out a million times the luminosity of our own.

The Earth gives you 5x10^14 square meters of radiator surface area for free, and there’s not enough methane in the ground to power the starship launches needed to build an equivalently sized radiator in space. What good is scalability that you can’t afford to use?

The comparison with fossil fuels is a bit odd because we aren’t going to run out of our planet’s ability to radiate away the energy that it’s already absorbing from the sun, whether we capture it or not. Sunlight generates the same heat whether it’s captured by a solar panel or a parking lot.

We could power our entire economy using the sunlight that falls onto the cropland currently used for growing the biofuels that provide 6% of our transport fuel. Not just current electricity, but also 100% electrification of the cars that are currently 6% fueled by that land. Plus electrification of all residential and commercial heating needs. Plus 100x growth of datacenter power usage. Plus a bit of extra electricity left over.

We still have a lot of easy scaling left to do. The scaling bottleneck is cost, not sunlight.