SpaceX details AI1 orbital data centers, supporting Nvidia and Google chips
Elon Musk has revealed that SpaceX's orbital AI data centers, dubbed AI1, will leverage existing Starlink V3 technologies and support Nvidia GB300 and Google TPUs. SpaceX aims to achieve 1 terawatt of power per year from these satellites and will scale production at its Bastrop, Texas "Gigasat" factory using Starship launches. The project aims to provide low-latency AI compute from orbit, with Google, Blue Origin, and Orbital also pursuing similar endeavors.
Key Takeaways
- AI1 satellites will feature 70-meter wingspans and provide peak power levels of 150 kilowatts.
- SpaceX is expanding its Bastrop, Texas factory to 11 million square feet to mass-produce orbital hardware.
- The roadmap targets 1 gigawatt of orbital power by late 2027, eventually scaling to 1 terawatt.
- Data centers will connect via onboard laser links and use Starlink's Ka- and Ku-band antennas for ground transmission.
Why It Matters
The shift toward orbital AI compute addresses the critical energy and cooling shortages currently bottlenecking terrestrial data center expansion. By relocating high-density compute like Nvidia GB300 racks to space, SpaceX bypasses Earth-bound grid constraints and water intensive cooling requirements. This creates a new competitive tier for hyperscalers like Google and Amazon, who must now decide between building terrestrial capacity or joining the orbital race via Project Suncatcher or Blue Origin’s Project Sunrise. Watch for the first AI1 volume production milestones at the Bastrop facility by Q4 2027 to signal if orbital AI moves beyond experimental status.
Additional Context
The orbital compute race is intensifying as terrestrial constraints mount. Per MLQ.ai (June 2026), SpaceX’s announcement comes just days before its planned June 13 IPO under the ticker SPCX, with the company seeking a valuation up to $2 trillion. The orbital initiative is positioned as a secondary growth engine alongside Starlink. Concurrently, Nvidia-backed Starcloud successfully operated an H100 GPU in orbit late last year, running Google's Gemma model to prove that data-center-grade hardware can survive cosmic radiation and thermal cycling (per CNBC, December 2025). While SpaceX targets a million-satellite constellation, competitors are filing their own massive orbital blueprints. Bezos’ Blue Origin filed FCC paperwork in March 2026 for "Project Sunrise," a 51,600-satellite network intended to offload water-intensive AI training from Earth (per Data Centre Magazine, March 2026). Google is also advancing "Project Suncatcher" in partnership with Planet Labs, with initial test launches scheduled for early 2027 to validate TPU performance in sun-synchronous orbits (per Cloud Computing News, June 2026). however, the industry faces significant skepticism regarding the economic and environmental costs. Despite Musk's claim that no "magic" is needed, orbital data centers may generate significantly more emissions than terrestrial equivalents due to frequent rocket launches and satellite reentry (per Scientific American, March 2026). Additionally, the physics of heat rejection remains a hurdle; the International Space Station currently requires nearly four times the radiator area per kilowatt compared to the specs SpaceX has proposed for the AI1 (per Tom's Hardware, June 2026).
Read full article at lightreading.com
Get this in your inbox → Subscribe
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source