Barclays: Orbital Data Center Costs Are 3x Ground-Based
Taylor Wilson
Barclays estimates orbital data centers cost roughly 3× more per megawatt than ground-based ones, and whether SpaceX's Starship can drive launch costs below $200/kg will determine if the concept becomes viable this decade.
Why move data centers into orbit?
Hyperscale cloud operators plan to spend roughly $800 billion on data center builds this year, but ground-side power, land, and grid access are tightening fast.
Of the 16 GW of new capacity planned this year, about half is already stalled — only 5 GW is under construction.
This means → space isn't suddenly cheap; ground-side bottlenecks are forcing the industry to look up.
What makes orbital data centers attractive?
Satellites in sun-synchronous orbit — a trajectory that keeps them in near-continuous sunlight — can tap constant solar power with no need for grid access, land permits, or water cooling.
Barclays analyst Brendan Lynch notes that orbital solar panels can generate up to 8× the energy of ground-based panels, thanks to uninterrupted sunlight and no atmospheric interference.
In plain terms = the three biggest headaches for ground data centers — where the power comes from, who approves the land, and whether there's enough water — all vanish in orbit.
Does the economics work?
Lynch estimates five-year build-and-operate costs at roughly $51 billion per GW in orbit versus $16 billion per GW on the ground — a gap of about 3× per megawatt.
Google's own analysis is more specific: orbital computing becomes economically viable only if launch costs fall below $200/kg by 2035.
SpaceX's Falcon Heavy currently runs at roughly $1,500/kg — still more than 7× above that threshold.
This means → technically imaginable, but economically still an order-of-magnitude cost compression away.
Why is Starship's timeline the core variable?
Whether launch costs can fall fast enough hinges on the commercialization of Starship — SpaceX's super-heavy, fully reusable rocket under development.
Lynch expects full Starship commercialization around 2027–2028, with true scale arriving by the end of this decade.
Reuters reports SpaceX plans to launch an orbital AI computing demo mission as early as late 2027 — a key proof-of-concept milestone.
Does this threaten existing data center companies?
Lynch states explicitly: over the next 10 years, orbital data centers pose no threat to traditional data center players such as DLR, EQIX, IRM, and AMT.
His view beyond that horizon: impact becomes harder to assess, but even if space data centers materialize, they would "likely complement, not replace" traditional deployments.
This reflects a market narrative running ahead of engineering reality — SpaceX's IPO pushed this concept from whiteboard to investable thesis, but commercial validation still needs years of flight records to answer the question.
Content is for reference only, not financial advice.