The huge wait times for turbine manufacturing I am talking about are for data centers, not general metro area power.
Quote from: Vultur on 10/29/2025 12:00 amThe huge wait times for turbine manufacturing I am talking about are for data centers, not general metro area power.It stands to reason that a turbine is a turbine is a turbine. So if you want to power your data center by a turbine plant, you take a number from the turbine line, which is as you say long. Now serving, 0-1-5. WHAT?!?!I don't know how long it takes to site and build a field in any of Earth's deserts. But given the operating parameters above, I'd expect everything about the project to be easier for a data center than it is for metro power.There needs to be some adjustment time, since the mad rush for AI power is relatively recent, but if it continues, I expect the sight of data centers surrounded by enormous PV fields out in the middle of nowhere to become common.
And for the same reason, I don't expect data centers in orbit beyond the novelty phase.
The great capacity of New Glenn should be talk more...and other rockets like Neutron, Terran-R, etc...
Quote from: meekGee on 10/29/2025 01:18 amQuote from: Vultur on 10/29/2025 12:00 amThe huge wait times for turbine manufacturing I am talking about are for data centers, not general metro area power.It stands to reason that a turbine is a turbine is a turbine. So if you want to power your data center by a turbine plant, you take a number from the turbine line, which is as you say long. Now serving, 0-1-5. WHAT?!?!I don't know how long it takes to site and build a field in any of Earth's deserts. But given the operating parameters above, I'd expect everything about the project to be easier for a data center than it is for metro power.There needs to be some adjustment time, since the mad rush for AI power is relatively recent, but if it continues, I expect the sight of data centers surrounded by enormous PV fields out in the middle of nowhere to become common.That makes sense, but then why are data center people still ordering turbines despite the wait times? Why don't they all just go to solar+batteries instead? I think there has to be some practical limitation that makes it not as easy as the top line cost per watt makes it sound.I mean, ideally solar kind of has to be the best power source for most things. There's just so much more energy available than anything else. But nonetheless, while solar is growing very rapidly, people still build lots of other things.Quote And for the same reason, I don't expect data centers in orbit beyond the novelty phase.I kind of don't either, because I don't really expect the current demand to continue long enough to really build it out. But I do hope the deployment of big low mass solar panels and radiators can get proved out first; those are enabling technologies for a lot of things.However, from the tweets above, it sounds like they are working from the assumption of basically unlimited future AI demand. Accepting that assumption for the sake of argument (though I don't actually believe it) I think there is a scale of energy use where moving to space makes sense or even is necessary; the question is just where that scale is, and whether anyone can afford to do stuff at that scale.
Either way, the value of sunshine in orbit can be quantified, and bounded from above, by assuming PV in orbit is free, and making some reasonable assumptions on the cost of terretrial, be it turbine or PV.
I would think that scaling PV industry on Earth is a lot easier than scaling manufacturing of orbital AI hardware.
I don't like assuming one side (orbital AI) is easy to scale, and at the same time assuming the other (terrestrial PV for example) is somehow naturally bound.
ApoStructura@ApoStructuraRocketLab is a cool company, but it isn’t a competitor to SpaceX.RL launches an order of magnitude less than SpaceX, and about 1000x less in terms of mass to orbit.Neutron is going to help but SpaceX is just that far ahead, and Starship is on the way.
To be viable for GW scale, Starcloud need $100/kg launch prices or lower. Neutron might get to $1000/kg. It’d need to grow a lot, and have a reusable upper stage, to reach $100/kg.
Falcon 9 is, internally, $1000/kg. It’s possible Neutron gets lower if they scale up a lot. But it’ll take a decade to amortize development, and RL is likely to still charge similar prices, $2000-3000/kg.
Philip Johnston@PhilipJohnst0n·I couldn’t be more excited to share that our @Starcloud_Inc1-1 satellite is targeted to 𝗹𝗮𝘂𝗻𝗰𝗵 𝗶𝗻 𝟮 𝗱𝗮𝘆𝘀, on November 2nd at 1:09 a.m. ET from Space Launch Complex 40 at Cape Canaveral Space Force Station in Florida as part of the Bandwagon-4 rideshare mission with @SpaceX. This is the first time anyone has flown an @NVIDIA H100 in space, which is orders of magnitude more powerful GPU compute than has ever been in space before. The launch webcast will be live-streamed on @X. Tune in! 🚀(Below are @ezrafeilden, @adi__oltean, and I at Cape Canaveral with the integrated Starcloud-1 satellite a few weeks ago!)
"That's not where we are," Gates said.But the billionaire philanthropist does think the current bubble is akin to the dot-com bubble of the late 1990s and early aughts, when several Internet-based companies turned out to be overvalued, resulting in a significant crash."In the end, something very profound happened. The world was very different," Gates said. "Some companies succeeded, but a lot of the companies were kind of me-too, fell behind, burning capital companies.
”Absolutely, there are a ton of these investments that will be dead ends," Gates added.
Philip Johnston@PhilipJohnst0nStarcloud-1 successfully launched!! 🚀🚀@Starcloud_Inc1, @ezrafeilden, @adi__olteanThanks for the ride @SpaceX! 🙏
Philip Johnston@PhilipJohnst0nI’ve never heard a more beautiful call out:“𝗦𝘁𝗮𝗿𝗰𝗹𝗼𝘂𝗱-𝟭 𝘀𝗲𝗽𝗮𝗿𝗮𝘁𝗶𝗼𝗻 𝗰𝗼𝗻𝗳𝗶𝗿𝗺𝗲𝗱” !!!!!!🚀🚀🚀🚀🚀🚀@Starcloud_Inc1, @ezrafeilden, @adi__oltean
Philip Johnston@PhilipJohnst0nThanks @GadiNBC for having me on @NBCNews to discuss putting the first @Nvidia AI chip in space with Starcloud_Inc1! 🚀
Y Combinator@ycombinatorStarcloud (@Starcloud_Inc1) recently made history by launching a satellite with an NVIDIA H100 into orbit — the first time a GPU that powerful has ever operated in space. It's the first step toward building AI data centers in orbit, powered by continuous sunlight and cooled by radiating heat into deep space.Their approach could one day rival the world's biggest data centers while using less energy, zero fresh water, and far lower emissions.YC's @aaron_epstein visited Starcloud's HQ, where co-founders @PhilipJohnst0n, @EzraFeilden, and @Adi_Oltean explained how they built a working prototype in just 15 months — and why big tech is racing to space for AI compute.
Quote from: Tywin on 10/28/2025 09:39 amThe great capacity of New Glenn should be talk more...and other rockets like Neutron, Terran-R, etc...Another reason that Phillip isn't interested in RLQuoteApoStructura@ApoStructuraRocketLab is a cool company, but it isn’t a competitor to SpaceX.RL launches an order of magnitude less than SpaceX, and about 1000x less in terms of mass to orbit.Neutron is going to help but SpaceX is just that far ahead, and Starship is on the way.https://x.com/ApoStructura/status/1983298258429718840
