McCalip's analysis finds orbit-based compute possible but costlier than terrestrial data centers, with LCOE near $891/MWh and mass around 22 million kg.
Science Team
Continue your reading
McCalip's Orbit Compute Framework: Space Power vs Earth Data Centers
A back-of-the-envelope take argues that computing in orbit is physically possible and could, under certain assumptions, compete with Earthbound data centers. The piece asks a straightforward, almost unglamorous question: why compute in orbit at all? McCalip frames it plainly: “Why compute in orbit? Why should a watt or a flop 250 miles up be more valuable than one on the surface?” The takeaway isn't that orbital data centers are a done deal, but that the economics aren't self-evident and deserve close scrutiny rather than hype.
Context helps explain why this matters. Terrestrial data centers have become energy and land-use behemoths, with rising power costs and cooling demands that constrain where they can sit and how much they can grow. In that light, orbiting some compute tasks might seem like a clever workaround if it unlocks advantages such as abundant sunlight for power, reduced land use, or resilience to weather on Earth. McCalip emphasizes that the analysis rests on first-principles modeling using publicly available data, with no proprietary inputs. He explicitly notes that the page is “built from publicly available information and first-principles modeling. No proprietary data,” a reminder that the argument is theoretical and meant to illuminate trade-offs rather than promise a short-term fix.
In his orbital scenario he lists Orbital Solar at about $31.2 billion, Satellite at $9.0 billion, Launch at $22.2 billion, Ops (about 1% per year) at $3.1 billion, and NRE plus replacement at $1.0 billion. The resulting metrics are cost per watt of $31.20 and an LCOE of roughly $891 per megawatt-hour, with a mass to LEO of about 22.2 million kilograms. By comparison, the terrestrial path centers on a total around $14.8 billion, with breakdowns for Power Gen, Electrical, Mechanical, Civil/Shell, Fit-out, and Fuel. That route yields a cost per watt of about $14.80 and an LCOE near $398 per MWh, with capex around $13.80 per watt. In short, the orbital route is far more expensive on a per-watt and per-hour basis in this schematic, and it carries a hefty mass and logistics burden to put the hardware into space.
These numbers highlight a broader point McCalip stresses: the orbital option isn't a sure thing. The upfront costs and launch risks are substantial, while the terrestrial path benefits from established infrastructure and lower per-watt upfront costs. But the analysis isn't guesswork. It highlights the fundamental trade-offs that would determine whether orbit-based compute could ever win on economics: launch costs, the energy balance of in-orbit power, system reliability in the radiation environment, data-link latency and bandwidth, and the long tail of maintenance and replacement in space. The takeaway isn't a guarantee of viability, but a framework for evaluating when and if orbit-centered data centers might become attractive.
Looking ahead, the piece invites careful follow-up work rather than premature conclusions. If rocket costs fall, if in-orbit maintenance becomes easier, or if demand for space-centric compute grows in applications such as deep-space mission science or latency-sensitive space assets, the balance could shift. The discussion also raises important questions about what we value in a data center: the cost of energy, the reliability of a given power source, the geography of disaster risk, and the logistics of sustaining a large fleet of spacecraft. McCalip’s disclaimer that this reflects his methodology and publicly available data anchors the piece as a thought experiment meant to provoke rigorous, independent analysis rather than declare a new industry trend.
For readers curious about the broader context, the economics of data centers and energy use is an area where institutions such as NASA, national labs, and universities contribute to the public conversation. The concept of space-based power and computing sits at the intersection of energy systems and space technology, a field with ongoing research and debate at space agencies and research centers. If you want to explore related sources on energy economics and space power, consider browsing authoritative sources on LCOE and data-center energy efficiency, such as the National Renewable Energy Laboratory’s explanations of LCOE and the U.S. Department of Energy’s data-center energy resources. For a broader scientific perspective, Nature and related scientific publishers provide ongoing coverage of energy systems and space technology research, while NASA’s own pages outline space power concepts and how they fit into future mission planning.
External sources you may find informative: