Welcome to Bedtime Astronomy. Explore the wonders of the cosmos with our soothing Bedtime Astronomie podcast. Each episode offers a gentle journey through the stars, planets, and beyond, perfect for unwinding after a long day. Let's travel through the mysteries of the universe as you drift off into a peaceful slumber under the night sky.
Okay, I want you to start by picturing something well, truly immense, the Sun. Our sources are telling us that the Sun, just doing its thing, pumps out more power than me. Get this, one hundred trillion times all the electricity humanity generates across the globe. I mean that number. It's not just big, it's almost impossible to really wrap
your head around, right, Yeah, that scale of energy. And yet down here on Earth or a modern world, especially our push for bigger and bigger AI artificial intelligence, it's kind of hitting a ceiling, a real physical limit based on how much energy we can actually feed these things now we cool them. We're burning through massive amounts of power just to train and run the really big machine learning systems, and that demand is just going up and up.
So what happens when a company like Google, I mean, a company that's all in on scaling AI, looks up and asks, well, what if the best place, maybe the only sustainable place to really scale this stuff isn't here. What if it's you know, up there in space, tap and right of that massive power source. Today we're doing a deep dive into a really bold idea, Google's project Suncatcher.
We've got this fascinating, super detailed research paper that outlines their technical plan for actually moving entire data centers, huge stacks of computers running their own custom chips in the low Earth orbit. It sounds like something straight out of sci fi, honestly, but when you look at the engineering details, they're surprisingly well grounded in tech. It's either here now or very close.
Yeah, And that grounding, that sort of technical seriousness, is what makes this more than just a cool thought experiment. It's real engineering because when we talk about AI right now, we're basically talking about an energy crisis that's just around the corner. Training the biggest models today, they already use more power than small cities. So projects on Ketcher it's a serious engineering first attempt to tackle what's fast becoming
a huge bottleneck for the whole tech industry. It's an infrastructure problem fundamentally, So for you listening in, we're going to try and get past that initial wow factor and really unpack the nuts and bolts of how Google actually thinks they can make this work. We've got three main areas we're going to break down today. First, the really huge, undeniable energy advantage you get in orbit and the specific
orbital mechanics they need. Second, the let's call it extreme engineering that's required for the communication and for keeping a whole data center flying together in a tight formation. And finally, something pretty surprising how well the hardware might actually hold up, and the razor thin economics that really decide if this whole thing makes sense in the next say decade or two.
Okay, let's start right there with energy, because that's the whole point, isn't it. That's the core problem projects suncatchers aiming at. Like you said, the power demands for large scale machine learning are just enormous, So why mess around building ever bigger solar farms down here where you've got clouds, nighttime, the atmosphere blocking sunlight land issues. Yeah, when you could just put the computers right next door to the power source itself.
It really boils down to efficiency, doesn't it. If energy is your biggest cost, you go where the energy is cheapest and most abundant, and the suncatch paper or their analysis it suggests that solar panels in orbit could be up to eight times more productive than the best ones we have down here on Earth. Wow, that eight x factor, that's the fundamental thing driving this whole project. It just completely changes the math before you even start talking about chips or rockets or anything else.
Eight times more productive, that's a staggering difference. Why why such a huge gap? Is it just because there's no atmosphere up there?
That's a big slice of it. Yeah, our atmosphere, you know, great for breathing, but it absorbs or reflects something like thirty percent of the Sun's energy right off the bat and that's on a clear day, not counting clouds or dust or anything. Up in LEO low Earth orbit, you're getting the raw deal, full spectrum, no filtering, pure sunlight. But there's actually another factor that's maybe even more powerful, especially when you think about cost, and that's the idea
of continuous collection. See down here, even in the sunniest desert on Earth, you lose half your operating time just because of night plus you've got seasons, bad weather. But in orbit, especially the orbit they're planning, these satellites are in almost constant sunlight.
Almost constant sunlight. Okay, And if you get rid of the night cycle, what big engineering headache just.
Vanishes storage batteries. It basically eliminates the biggest, heaviest, and frankly most expensive part of large scale solar power on Earth, which is storing enough energy to last through the night. If you're generating power pretty much twenty four to seven, you just don't need those massive, heavy, costly battery banks that are absolutely vital for any serious solar project down here.
Ah right, that makes sense.
Heavy, incredibly heavy. Think about the mass budget for launch. Lithium ion batteries add a lot of weight. To power a data center on Earth through the night, you'd need enormous battery arrays adds mass, adds volume, adds cost in space. If you need far less battery mass, your satellite is lighter, and that directly lowers the complexity and crucially the cost to launch the whole thing, So that AIGHTX efficiency isn't
just about getting more power out. It's also about needing left heavy stuff like batteries sent up there in the first place. It's minimizing that dead weight.
Okay, so Project Suncatcher isn't just about grabbing more sunlight. It's using a really specific orbital trick to get that continuous power. So let's talk about what this thing actually looks like. What's the architecture. Is it one giant space station?
No, not monolithic at all. The concept is actually a distributed constellation. Think lots and lots of separate satellites. Each one would have its own solar panels, its own processors, specifically Google's own ships, their TPUs, and they'd all be connected together with super fast laser links, optical links. And the orbit they chose that's really the key piece of engineering here. They've zero in on a sun synchronous.
Low Earth orbit l e O Sun synchronous ALIO.
Yeah, sometimes called SSO, and choosing that specific orbit. It's not random at all. It's this really clever bit of orbital mechanics, a kind of geometric trick designed purely to keep the sunlight hitting those panels almost all the time.
Okay, for those of us listening who hear Sun synchronous and maybe you just picture it following the Sun around. Can you break down the orbital magic there? What makes SSO so good for power?
Sure? So, an SSO is an orbit that goes almost over the poles, a near polar orbit, and the magic, as you put it, is how the plane of that orbit rotates. The orbit itself actually precesses. It sort of turns around the Earth at the exact same speed that the Earth goes around the Sun, which is about one degree per day. Okay, So the geometry works out such that the satellite always crosses the equator at the same local solar time every day.
Ah Okay, so the Sun is always hitting it from roughly the same direction relative to.
Its pass exactly. The satellite keeps a near constant angle relative to the Sun, so the solar panels can be oriented to be almost perfectly perpendicular to the sunlight for most of the orbit, maximizing collection. And the big payoff is that the satellite hardly ever goes into Earth's shadow. If it does, the eclipse is super short, maybe just a few minutes. Compare that to a typical l EO satellite, which might spend thirty or forty minutes in darkness every
single orbit difference, huge difference. This specific orbit choice guarantees that almost constant sunlight, which then lets them drastically minimize how many batteries they need onboard. It's really the sweet spot maximum sun exposure for power, but still in low orbit, which helps keep the communication latency back down to Earth somewhat reasonable.
Okay, that perfect energy supply, that constant sunlight. It leads us straight into the second, really big challenge, maybe the hardest one. Taking a static building some data center on the ground and turning it into a swarm of individual satellites flying together information that demands coordination on a scale that just sounds well, incredibly difficult.
Yeah, and the coordination isn't just a nice to have, It's an absolute requirement, and it's driven entirely by the need for fast data links between the satellites, getting that data center level of performance the kind you need for these giant AI models. It absolutely depends on the satellites flying and what the paper calls extremely tight formation.
Extremely tight Why why so close as it just so the lasers don't miss each other.
That's definitely part of it. The pointing accuracy but the main reason comes back to the data link speed and the latency. We'll get more into the links themselves in a bit, but think about how a huge AI model works. It's splitting up the job across potentially thousands of processors simultaneously. All those processors need to talk to each other, share results, coordinate almost instantly like.
Components inside a single computer.
Basically pretty much in a data center on Earth, those processors might be connected by a few feet of fiber optic cable, really short distances, giving you microsecond latency incredibly high bandwidth. To get anywhere close to that kind of low latency in space, you just can't have the satellites drifting far apart. The paper specifies they need to be separated by kilometers or less.
A kilometer or less, okay, even a kilometer in space. When you're talking data links, that still sounds like a long way for data compared to inside a server rack.
It is, it absolutely is, And you can't beat the speed of light.
Right.
Even though light travels faster in vacuum than in fiber, that kilometer distance still introduces a delay you don't have on Earth. So they have to design the whole system the network protocols, the way tasks are distributed. Specifically to handle these slightly longer latencies, The whole constellation has to function like one single massive coordinated computer cluster, even though it's physically spread out over potentially you know, several miles.
And the altitude they're aiming for around six hundred fifty kilometers. That adds another complication, doesn't it. It's LAO, sure, but it's not that high. There's still some atmosphere, some drag.
Precisely, this isn't way out in geo geostationary orbit, where drab is basically zero down at six hundred and fifty kilometers, Especially when you're trying to keep potentially hundreds of these things packed so closely together, maybe hundreds of meters apart or less. Even tiny forces become a.
Huge problem, like what kind of forces.
Well, the residual atmospheric molecules, for one, Even though the air is incredibly thin up there, it's not a perfect vacuum. Hit a slightly denser patch, and it creates drag. If that drag affects one satellite's big solar panel slightly differently than its neighbors.
It could nudge them out of alignment almost immediately and break those laser.
Lis exactly, break the links, and the whole cluster stops working as a single unit. Plus you've got subtle things like the fact that Earth's gravity isn't perfectly uniform. All these tiny perturbations add up when you need such precise relative positioning.
So if the forces are tiny but complex and always changing, how did Google figure out if they could actually solve this, this problem of keeping a distributed data center flying in formation seems almost impossible.
Well, they didn't just guess. This is where the serious engineering comes in again. They built some really sophisticated physics simulations to model it. They wanted to analyze the stability of these tight formations over the long haul, over the whole mission lifetime.
What did the simulations look at?
Specifically, they modeled two main threats. First, those tiny irregular gravitational tugs from Earth not being a perfect sphere, what they're called the non gravitational field effects. And second, maybe more importantly in Lo, that atmospheric drag, which, like we said, is small at six hundred and fifty kilometers, but it's there and it changes depending on things like solar activity,
puffing up the atmosphere. Even tiny differences in drag between adjacent satellites over time could cause them to drift apart significantly if you don't correct for it.
Okay, that makes sense. You'd think hitting a slightly denser bit of air would slow one satellite down relative to its neighbor. Wouldn't that mean you need constant, powerful thruster burns to keep nudging them back into place. That sounds as a lot of fuel.
You'd think so, right, But the conclusion from their models was actually surprisingly positive, which is good news for the project's lifespan. The simulations indicated that they should only need modest station keeping maneuvers to keep the formation stable over time.
Modest that's the keyword there.
It really is, because if they needed constant high thrust chemical rockets firing, they'd burn through propellant incredibly fast. That would limit the mission life drastically, and the whole economic argument kind of falls apart.
So modest maneuvers probably means they're thinking about using things like electric propulsion, ion drives, hall thrusters, stuff that SIPs fuel but provides tiny pushes over long periods.
That's almost certainly the implication Yeah, low thrust, high efficiency propulsion is perfect for this kind of longeration fine adjustment job. They can provide continue uk US really small corrections for years without needing massive fuel tanks, which means the mass needed for station keeping stays manageable within the overall satellite budget, and that helps keep the launch cost down. It all connects. The source material actually mentions NASA's Stereo Observatory spacecraft as
a kind of parallel. Those probes needed really precise orbital control at relatively low altitudes to maintain their view of the Sun. The fact that Google's modeling suggests only modest maneuvers are needed implies that while the formation is tight, the natural orbital dynamics aren't totally fighting them. They can use these efficient, low thrust systems to nudge things back
into place. It means that huge energy advantage they get from the sun synchronous orbit isn't immediately canceled out by needing tons of fuel just to hold the constellation together.
Okay, all right, So we've got the massive power source, sorted things of the orbit, and we've got a plausible, simulation backed path to keeping the satellites flying to gather in formation using efficient thrusters. Now the third huge piece communication because if you can't shift data between all those processes at blinding speed was really low latency, then you don't have a data center. You just have a bunch of well solar powered computers sloating near each other. Not useful for big AI.
Yeah, this communication challenge, it's truly the make or break piece for Suncatcher. We talked about AI workloads needing high bandwidth and low latency because the tasks get split up across maybe thousands of processors. If these connections between the processors are slow, the whole calculation grinds to a halt. It doesn't matter how fast the individual chips are. The system runs at the speed of its slowest link.
Right.
So, to really perform like a data center on Earth where data zips around on fiber optic cables at tens, maybe hundreds of gigabits per second, these links in space have to hit similar numbers across that kilometer or so gap we talked about, and Google set the bar incredibly high here. The goal they state is links supporting tens of terabits per second between satellite.
Tens of terribus per seconds that's astronomical way beyond current inner satellite links, isn't it.
Oh yeah, orders of magnitude faster. So how do you possibly achieve that kind of speed across kilometers of empty space, and crucially without using huge power hungry radio antennas. The answer is lasers. They're relying entirely on laser based optical communication. That's the key technology here.
Lasers. Okay, why lasers instead of radio?
Well, unlike radio waves which tend to spread out over distance and need a lot of power to keep the signal strong, lasers can focus light into an incredibly tight, narrow beam think laser pointer versus a floodlight. That tight focus means much less power gets wasted and you can pack way more information, much higher throughput into that beam.
We've seen demos of this kind of thing before, right, leaser cons in space.
Yeah.
Absolutely. The sources mention the opal US experiment on the International Space Station, for example, optical payload for lasercom science that proved you could beam high bandwidth data using less pretty effectively. But the challenge here is taking that proof of concept and scaling it up massively, like by a factor of ten or twenty maybe more to get to those tens of terrabits.
Okay, so the basic tech exists, but going from say gigabits or maybe single terabt up to tens of terrabits, that's a giant leap. How do they plan to squeeze so much more data down the same laser plate.
They're planning to use some really advanced multiplexing techniques, basically finding clever ways to pack more separate data channels into the same physical path. The paper details two main methods working.
Together multiplexing like splitting the beam sort of.
The first is called dense wavelength division multiplexing or DWDM. This is already standard practice in long haul fiber optics on Earth. Imagine the laser beam is like a single highway. DWDM is like painting say thirty or forty different color lanes onto that same highway. Each color, each specific wavelength of light carries its own independent stream of data.
Ah. Okay, so multiple data streams riding on different colors of light within the same beam. That multiplies the capacity right there exactly.
It leverages the spectrum really efficiently. But then they add a second layer on top of that.
Okay, what's the second layer.
That's spatial multiplexing technology. So if d do UDM gives you multiple lanes on one highway, spatial multiplexing is like building several parallel highways right next to each other. It means sending multiple separate laser beams between the two satellites simultaneously, all aimed very precisely at an array of sensitive receivers on the other end.
Wow, okay, so multiple beams and each beam is carrying multiple colors or wavelengths of data precisely.
You combine DDM and spatial multiplexing multiple colors times multiple beams. That's how they aim to multiply the data rate so dramatically and hit that collective target of tens of terabits per second.
That sounds incredibly complex keeping all those beams and colors aligned across a kilometer of space. Is this still just theory or have they actually tested any of.
This is important. They have done some concrete validation, at least on a lab bench, which is crucial right to show the physics works. The research teams set up a demonstration combining these approaches, and they successfully transmitted data at a total rate of one point six terabits per second.
Okay, one point six tvps. It's still a long way from tens, but it's well, it's not zero. It shows the combined technique actually functions exactly.
It proves the principle. And what's really interesting here is how this lab success ties straight back to that need for tight formation flying we discussed earlier keeping the satellites separated by kilometers or less. That tight spacing it isn't just for orbital stability. It's absolutely essential for making sure these complex laser links can actually be scaled up and
work reliably. The closer the satellites are, the easier it is to maintain that pinpoint accuracy needed for multiple laser beams carrying multiple wavelengths to hit tiny receivers without the signal degrading or getting lost.
Ah I see. So maintaining that close distance is basically what allows the fancy multiplexing tricks to work at their full potential. It protects the signal quality for those super high data rates. The whole system is interconnected.
Totally interconnected. The formation flying enables the high speed links, and the need for high speed links dictates the type formation. One can't work without the other. If the formation drifts too far apart, the beams misalign, the data rate crashes, and your space data center just isn't a data center anymore. It's just separate computers again.
Okay, let's switch gears then to the computers themselves, the processors. You can have all the power in the universe, perfect formation, amazing laser links. But if the actual chips doing the work, the TPUs, get fried by space radiation on day one, then the whole project is obviously useless. And the usual thinking, right is that regular commercial computer chifts just can't handle
the radiation environment in space. You normally need special rad hardened components, which are expensive and usually slower.
That has definitely been the big hurdle for using standard off the shelf hardware in space for a long time. But what's really striking in the Sunketcher page is what they found when they tested Googles on chips. It suggests maybe that old constraint isn't quite as rigid as we thought,
at least for their specific architecture. They took their Trillium V six E Cloud TPU, that's their custom chip designed specifically for AI and machine learning, and they blasted it with radiation to see how it held up, and the results they're actually much better than even the researchers themselves seemed to expect better.
In what way, how do you measure holding up against radiation? We're talking about accumulated damage over time, right, Yeah, and also maybe sudden failures from like a direct hit bioparticle.
Yeah, both are concerns. The study focused heavily on the cumulative dose tolerance, which is measured in a unit called rads. It's basically how much radiation energy gets absorbed over time. So for the planned five year mission in that specific sun synchronous l EO, they calculated the expected total radiation dose that chips would receive even with some shielding. The number they planned for their baseline was seven hundred and fifty rads.
Seven hundred and fifty rads over five years.
Now here's the key finding. The TPU chips themselves, the core processors could actually withstand a cumulative dose almost three times higher than that. They kept working properly well past seven hundred and fifty rads, significantly past it before they started showing any major problem.
Wow, three times the expected dose. That's a huge margin. Does that mean that maybe the way modern chips are built really small and dense, actually helps protect them somehow more than older chips.
That seems to be exactly the implication. Yeah, modern ship manufacturing with incredibly tiny transistors packed cluickally together smaller nodes, it seems they might be inherently less likely to be damaged by certain kinds of radiation, just less physical target area for the radiation to hit.
Perhaps.
If this holds true, it's a massive deal Economically. It means they might be able to use largely commercial grade processors, which are produced in huge volumes and are relatively cheap, instead of relying solely on expensive, low volume, specially hardened space components that could drastically lower the cost per satellite.
Really, not every part of the system is equally tough in a complex chip, Aren't things like the memory usually more delicate.
That's a very good point, and yes, the testing showed exactly that. While the main processing units were surprisingly robust against the total dose, the high bandwidth memory or HBM systems turned out to be the most sensitive part HBM.
That's the really fast memory right next to the process, or crucial for feeding it data quickly for AI tasks.
Yeah, exactly. It's absolutely vital. If the HBM fails or gets corrupted too often, the whole AI calculation grinds to a halt, no matter how tough the processor is.
So, how did the HBM fare in the tests? What was its breaking point?
Well, even the weak link, the HBM did surprisingly well. It only started showing significant issues related to the cumulative radiation dose after it had absorbed about two killer ads.
Two killer ads. That's two thousand rats.
Yep, two thousand rads. And remember the expected dose for the entire five year emission with shielding was only seven hundred and fifty rads.
So the most sensitive part lasted well over double the expected mission dose before failing.
Correct more than double, which strongly suggests that fundamentally, the high performance hardware Google is using is basically tough enough for that ILIO environment, assuming they designed the shielding correctly. This finding really tackles what many space engineers would probably point to as the biggest technical show stopper and cost driver for putting commercial style data centers in orbit. It suggests the hardware might actually survive.
Okay, this whole thing, Project Suncatcher, it's just an amazing combination of well energy physics, clever orbits, laser engineering, surprisingly tough hardware. But like we said at the start, Google's a business. This is a gigantic infrastructure play. It only actually happens if it makes financial sense. Right, When does putting computers in space actually become cheaper than just building another data center down here? What's the bottom line?
Yeah, that's the multi billion dollar question, isn't it? And the paper is refreshingly honest about it. The entire financial viable ability, the whole economic case for Suncatcher, it depends heavily on launch costs continuing their decline. If the cost per kilogram to get stuff out of Earth gravity well doesn't drop dramatically, then this whole project, no matter how cool the tech is, just stays on the drawing board. It's completely unaffordable. Right.
Launch cost is everything in space.
It's the biggest upfront cost by far. You have to lift all that hardware, the satellites, the solar panels, the lasers, the processors, the fuel for station keeping all the way into orbit.
Just to give people a sense of scale, what does it roughly cost today to launch a kilogram to l EO using current.
Rockets ballpark figures. Even with modern reasonable rockets coming online, you're probably still looking somewhere between say, fifteen hundred dollars and maybe three thousand dollars per kilogram on the commercial market,
maybe a bit less for very large customers. Now that's way way cheaper than the Space Shuttle air obviously, but launching the tens, maybe hundreds of tons of hardware needed for a constellation like Suncatcher at those prices, the upfront cost is just astronomical, completely non viable.
So what's the magic number? Then? What did Google's analysis say launch costs need to fall to for this whole thing to start looking economical compared to the energy.
Savings Their analysis pinpoints are really critical target price launch costs need to get below two hundred dollars per.
Kilogram two hundred dollars akilo.
Wow, yeah, two hundred dollars per kilogram. That's the threshold where, according to their map, the enormous cost of building, launching, and operating this space infrastructure starts to become economically justifiable when weighed against the huge continuous savings from getting free constant eight x more efficient solar power.
Two hundred dollars akilo is that's incredibly ambitious. That's not just incrementally cheaper rockets. That basically requires the full dream of super heavy, rapidly reusable launch vehicles to become reality, right like SpaceX's Starship working at full cadence or similar systems.
Absolutely, they are essentially making a long term strategic bet on that specific future for the launch industry, a future defined by full and rapid reusability and a very high flight rate. Their timeline reflects this. They project that this kind of price point sub two hundred kilograms might become achievable sometime in the mid twenty thirties, assuming you know, continued progress and success in developing these next generation launch system.
The mid twenty thirties. Okay, so this isn't something they expect to build next year. It's a decade plus strategic vision.
Definitely, it requires the whole commercial space ecosystem, especially launch, to mature significantly. And that two hundred dollar kilogram target probably also implicitly relies on the fact that because of the ad X energy efficiency and constant sunlight, the satellites themselves can be lighter, smaller solar arrays, way fewer batteries than if they had to launch the equivalent power generation
and storage capacity needed on Earth. So the total mass they need to launch is less to begin with.
Okay, so let's assume they hit that target. Let's say launch costs do get down to two hundred dollars per kilogram by the mid twenty thirties. What's the final economic verdict? Then? How does space compare to Earth?
This This is really the punchline that ties the whole vision together. At that target price point, the analysis concludes that launching and operating a space based data center becomes roughly comparable in cost to just the energy expenses of running an equivalent AI facility on Earth.
We say that again, the total cost of building and launching the space system equals just the electricity bill for the Earth based one.
Roughly comparable. Yeah, think about how huge that is. The massive capital cost of designing, building, launching, and maintaining this incredibly complex space constellation becomes similar in magnitude to just paying the ongoing power bill for a state of the art AI data center down here.
So basically the biggest operational cost on Earth, that relentless massive electricity drogic gets.
Utralized, pretty much neutralized yeah, it completely flips the usual cost calculation down here. Energy is the huge ongoing operational expense that kills you some Catcher frames. The big upfront launch cost is the main hurdle. But once you clear that hurdle, assuming cost drop enough, the free can continuous energy from the Sun provides this massive long term economic advantage.
It offsets what would have been your biggest terrestrial operating cost. Okay, let's try and quickly recap the main pillars of what we've dug into here with projects on Catcher, we're looking at this really radical but technically detailed plan to essentially cut the cord between future AI growth and the limits of Earth's energy grid. The technical plan is super specific.
We've got that huge energy advantage, solar panels maybe eight times more effective, and that carefully chosen sun synchronous LAO. We've talked about how they think they can get data center speeds between satellites using lasers with fancy techniques like DWDM and spatial multiplexing, but that relies heavily on keeping the satellites in that really tight formation less than a kilometer apart, using those modest efficient station keeping thrusters and
then there was the surprising finding about the hardware. Google's own TPUs seemed tough enough to handle the elo radiation environment, potentially lasting almost three times the expected dose for a five year mission. Even the more sensitive memory lasted twice aid as long as needed.
Right underpinning absolutely everything is the economics. It all hinges on launch costs dropping dramatically below that two hundred dollars per kilogram mark. Google thinks that's possible by the mid twenty thirties, and if they hit that number, the whole system starts to look financially competitive, purely based on saving the massive energy costs heeding cur on Earth. So, stepping back, what does this all really mean in the bigger picture
of technology of infrastructure? This isn't just putting servers somewhere else slightly inconvenient. It feels like a fundamental shift. If you can actually decouple massive computing power from the physical limits of our terrestrial power grids, which right now is maybe the single biggest thing holding back AI scaling, then you're not just getting a bit more processing power. You're potentially unlocking a whole new scale of AI development, aren't you.
It's about creating a tech base that isn't tied down by how many power plants we can build or where we can build them, or the politics around energy.
That really leads to the final big question. Doesn't it something for you, the listener, to really chew on after hearing all this, If the AI models we have today, which are still relatively limited in the grand scheme of things, if they are already straining power grids and consuming frankly enormous amounts of energy, what scale of machine learning? What
kinds of problems could we tackle? What level of intelligence could we potentially build if the energy source was effectively unlimited, tapping into something that's literally one hundred trillion times greater than our current global capacity, and the only real bottleneck left becomes the cost of getting the hardware up there. It's the kind of future scale project Suncatcher is aiming for. That's the potential transformation they're trying to engineers
Us