Making Clavius MoonBase for Real

SpaceX will make a large moonbase by 2036 as part of the industrialization needed for scaling AI data centers in space. SpaceX will need to make mining and factories on the moon in order to increase space based AI satellites from 100-1000 Gigawatts per year to 10 to 1000 terawatts per year. 10 Terawatts would be about 100 million tons per year of satellites. Clavius Base from 2001: A Space Odyssey is depicted as one of the largest and most advanced permanent U.S. lunar outposts in fiction. It houses 1,700 personnel (1,100 men and 600 women, all highly trained scientists, engineers, and technicians). The novel explicitly calls it as a miniature world in itself and a self-sustaining closed system that can produce all necessities of life (oxygen, water, food, etc.) from local lunar rocks via crushing, heating, and chemical processing, with a large underground “hothouse” for mutated plants and algae-based food systems.

Screenshot

Size and Layout
The base lies inside the real Clavius crater (≈225 km / 140 miles across; novel rounds to 150 miles). The vast majority of the complex is underground (excavated from solid rock or compacted lava) for radiation shielding, micrometeoroid protection, and thermal stability. Boring Company will need to send tunneling machines with SpaceX.

Surface elements include:
– A large central landing pad/dome (petal-like doors that open like an iris).
– An enormous elevator platform that lowers entire spacecraft (the spherical Aries 1B translunar shuttle, roughly 15–20+ m in diameter based on film models and visuals) dozens of meters into a cavernous underground docking bay.
– Ancillary structures like control towers, support buildings, and access points.

The footprint appears on the order of hundreds of meters to a couple of kilometers across (central pad ~100–200 m diameter, with radial/arc-shaped surrounding modules and infrastructure). Living quarters use standardized modular one-man rooms (≈60 square feet or 13.6 m³ each). They are arranged in multi-level underground complexes with shared workshops, labs, offices, computer centers, garages, kitchens, generators, and food-processing plants. It feels like a small underground city or military-style hardened facility repurposed for peaceful science.

Tons of Material to Build It (Movie-Era Analysis)
No official or precise figure exists in the film, novel, production notes, or analyses—construction details are background flavor.

Excavation and bulk shielding are mostly local lunar regolith/rock (piled or used in place), keeping structural mass down compared to fully imported habitats.

All pressurized modules, life-support systems, power plants, chemical processors, hothouses, vehicles, computers (1960s-era heavy mainframes), labs, food systems, and outfitting would be shipped from Earth.

Rough order-of-magnitude estimate (scaling from modern lunar base studies for 100–2,000+ crew, adjusted for the novel’s modular + underground approach and 1960s tech): 50,000–250,000+ metric tons total delivered mass over years of construction.

This factors in:
Habitable volume: Living modules alone ≈23,000 m³; full complex (labs, commons, agri, infrastructure) likely 200,000–500,000+ m³ pressurized.
Mass density: 200–500+ kg/m³ for pressure vessels, outfitting, and heavy equipment (higher than lightweight inflatables due to robust 1960s designs).
Power (likely nuclear or large solar), full recycling/chemical plants for self-sufficiency, mining gear, shuttles/rovers, and spares for 1,700 people.
This aligns with broader lunar base plans that have been proposed.

SpaceX, AI Data Centers, Mass Drivers, Lunar Solar/Satellite Production, and Teslabots

SpaceX scenario with Starship-class lunar landers (200-ton payload to surface capability with V4 Starship orbital refueling), heavy ISRU industrialization, mass drivers (electromagnetic catapults for efficient regolith/solar-panel/satellite launching to lunar orbit or beyond, reducing Earth dependency), on-Moon factories producing solar panels and comms satellites from regolith-derived silicon/metals/aluminum, AI data centers (for swarm control, optimization, simulation, and low-latency robot oversight—Moon’s cold vacuum and solar power are ideal), and Tesla Optimus bots (humanoid robots for labor).

Bootstrapping, Starship and Optimus changes everything

Initial seed payloads establish power, basic processors/sinterers/3D printers (for regolith mooncrete/bricks), smelters, initial robot fleet, and starter habitats. Robots + AI then exponentially scale production: mine/excavate, sinter structures, fabricate solar arrays, assemble satellites, build more bots/factories, and eventually mass-driver components. Local output covers 80–95%+ of bulk mass (habitats, shielding, power, infrastructure). Only high-tech seeds (advanced electronics, specialty alloys, biology starters, precision tools) stay Earth-imported.

Number of 200-ton landed payloads: 20–100 total (4,000–20,000 tons Earth material) spread over 5–10+ years.

Phase 1 (bootstrap, first 1–3 years): 10–30 landings for core industrial seed (nuclear/solar power plants, 1,000+ initial bots, regolith processors, printer farms, starter hab modules, chemical plants, AI compute racks).

Phase 2 (ramp-up): Additional 10–70 landings as needed for specialty gear while local production (solar farms, mass drivers, bot factories) takes over excavation, module fab, and expansion to full 1,700-person scale. Mass drivers accelerate regolith handling and could even export excess solar panels/sats to support orbital assets or further bootstrapping.
This is far lower than a no-ISRU build (orders of magnitude mass savings) and consistent with modern ISRU studies (some concepts bootstrap large bases with less 10,000 tons seed mass via exponential robot replication). SpaceX does not need to minimize the construction seed. Mass production of Starships that are funded by AI data centers will enable a speed focused program.

Teslabots and other gear

Teslabots/Optimus: Start with 200–2,000 advanced units sent in early payloads (for setup, mining, assembly). On-Moon factories (once solar/power and metalworking online) produce thousands more rapidly. Operational workforce: 5,000–20,000+ bots for continuous 24/7 labor (excavation, construction, maintenance, farming, logistics) in 1/6 g—far more efficient and safer than humans. AI data centers orchestrate swarms for precision tasks.

Autonomous heavy excavators/miners (a few dozen initial, then locally built), large-scale sinterers/3D printers, chemical refineries, vertical/agri domes (or expanded hothouses), mass-driver rail systems, satellite assembly lines, redundant power storage, rovers/haul trucks, and backup life-support.

3 thoughts on “Making Clavius MoonBase for Real”

  1. By the time they are up to speed through automation, why have any people? AI controlled bot swarms will do a far better job than any meat machines.

    • I think there’s a viable tourist angle. Much more realistic than people wanting to move to Mars permanently, at least anytime soon….

  2. 250,000 tons. 1,250 deliveries. So, let’s say 6000 Starship launches at $50M per launch.

    That’s about $300B and does not include the price of cargo.

    $50B for bootstrap with manufacturing. $50B more for a mass driver.

Comments are closed.