Etiido Uko is a news contributor for Tom's Hardware covering the latest updates in big tech and the PC industry. He is a mechanical engineer and senior technical writer with over nine years of experience in documentation and reporting. He is deeply passionate about all things engineering and technology, and is an expert in gadgets, manufacturing, robotics, automotive, and aerospace. ","collapsible":{"enabled":true,"maxHeight":250,"readMoreText":"Read more","readLessText":"Read less"}}), "https://slice.vanilla.futurecdn.net/13-4-25/js/authorBio.js"); } else { console.error('%c FTE ','background: #9306F9; color: #ffffff','no lazy slice hydration function available'); } Etiido Uko Social Links Navigation News Contributor Etiido Uko is a news contributor for Tom's Hardware covering the latest updates in big tech and the PC industry. He is a mechanical engineer and senior technical writer with over nine years of experience in documentation and reporting. He is deeply passionate about all things engineering and technology, and is an expert in gadgets, manufacturing, robotics, automotive, and aerospace.
Zaranthos The patents have some merit but they can't really restrict the natural evolution of satellite placement or the natural progression of trying to make low latency or orbital efficiency of data centers better. Companies like SpaceX would likely develop something similar without running afoul of these patents if the logistics of lower and higher orbit synergy made sense. Ultimately a lot of this is just the likely evolution of orbital data center progression especially once we connect AI to space station and moon base operations. I'm not sure how much value it has for terrestrial AI data center operations over low earth orbit networks that likely interconnect with laser speed systems anyway. Adding more latency with higher orbit operations has some trade offs when coordinating back on Earth. Reply
re3eyul the 3l0n will be there before Intel figures out how to make Leo , already got a factory doing Starship payload sized orbital compute modules . Reply
TerryLaze re3eyul said: the 3l0n will be there before Intel figures out how to make Leo , already got a factory doing Starship payload sized orbital compute modules . If you mean this one, it's not even started construction yet, and also it's going to use/be an intel FAB. https://www.tomshardware.com/tech-industry/semiconductors/terafab-starts-to-take-shape-100-million-square-feet-of-manufacturing-space-and-usd16-8b-initial-capital-investment Reply
chaos215bar2 A patent does not a "proposed orbital data center" make. The economics of the whole idea are simply ludicrous. Datacenters are big, heavy, power hungry machines that need cooling, regular maintenance, and upgrades to remain competitive. Literally none of that gets easier or cheaper in space. Reply
usertests chaos215bar2 said: Datacenters are big, heavy, power hungry machines that need cooling, regular maintenance, and upgrades to remain competitive. Literally none of that gets easier or cheaper in space. Even though Nvidia and AMD want to release new accelerators on a near yearly cadence (these roadmaps are slipping in practice), I think the rate of improvement could be slowing down by a lot. The move to support lower precision formats was like a free lunch that couldn't last forever. Yearly upgrades won't make sense for most users. Big users that are buying whatever they can will gradually phase out the old as they buy the new ones. You can't do any maintenance, so you have to make it robust and failure tolerant in the first place. It will last a few years before being deorbited to burn up, talk about planned obsolescence. Cooling in space is simply balancing an equation. It has to radiate away the energy that it uses, around 100-150 kW. So the satellites have to be relatively large, and fit in a payload bay using an unfolding design. It's a solvable engineering problem of a scale already accomplished by the International Space Station. They will put as many GPUs/accelerators on there as can be supported. If Starship becomes operational in a fully reusable configuration, it will be able to lift >100 metric tons to orbit at a relatively low cost, much faster than datacenters can be constructed on the ground. The supply chain to build a lot of these and lower costs is already there, since the satellites will be based on Starlink. The sun-synchronous orbits targeted will power the satellites nearly continuously. I won't claim the economics of this make sense, because it's entirely possible the numbers won't add up, or Starship will experience another string of failures and not be ready in time, or the AI bubble will pop and the AI compute demand won't materialize. I don't think it's impossible for this to make sense, but it's obviously risky. On the Starship front, they have gotten the ship suborbital (by choice), and released Starlink v3 payloads. The booster is experiencing a lot of engine failures. Both need to be reliable and reflown regularly to lower the launch costs. Reply
SmokyBarnable Zaranthos said: Adding more latency with higher orbit operations has some trade offs when coordinating back on Earth. Right. The distance from LEO to earth is much less than the distance from LEO to MEO much less to GEO, so I don’t understand why communicating with terrestrial is so horrible, at least as explained by this article. Reply
Key considerations
- Investor positioning can change fast
- Volatility remains possible near catalysts
- Macro rates and liquidity can dominate flows
Reference reading
- https://www.tomshardware.com/tech-industry/space/SPONSORED_LINK_URL
- https://www.tomshardware.com/tech-industry/space/intels-proposed-orbital-data-centers-would-manage-thousands-of-simple-leo-satellites-two-tier-network-puts-the-brains-of-satellite-constellations-in-higher-orbit#main
- https://www.tomshardware.com/membership
- NVIDIA Vera Rubin Driving Performance Per Watt, Lowest Token Cost for Partners Worldwide
- $4,429 order for a ROG Astral RTX 5090 cancelled by Nvidia due to a 'late' price increase, with Asus blamed — marketplace buyer refunded after immediate $500 in
- NASA modded space station's laptops so everyone could use the same charger — standardizing ISS chargers eliminated useless weight and reduced failure points
- NASA modded space station's laptops so everyone could use the same charger — standardizing ISS chargers eliminated useless weight and reduced failure points
- Pulsar Feinmann F01 Noctua Edition Review: Extra cool
Informational only. No financial advice. Do your own research.