“Do not leave for fear of an AI bubble”: why orbital data centres matter for investing (Lee Kyung-jun)
An independent exercise, not a statement by the video creator or verification of every claim in the video.
Abundant sunlight does not eliminate radiator area
Suppose a hypothetical computing device rejects 1MW of waste heat to space in steady state. Assume a radiator temperature of 300K, emissivity of 0.9 and one face exposed to space, excluding solar and terrestrial heat inputs and other heat-rejection paths. Using Q=εσAT⁴ gives approximately 413W/m² and an effective radiating area of 1000000÷413≈2420m². This is neither solar-panel area nor the spacecraft’s overall size. It shows why power availability and waste-heat rejection need separate calculations before radiator mass, deployment, heat transport and launch costs enter an economic assessment.
Check for yourself
- Have you distinguished computing power from waste heat and specified radiator temperature, emissivity and effective area?
- Does the actual design include incoming solar and terrestrial heat and radiator view and shadow conditions?
- Have you kept prototype results separate from evidence on large-scale lifetime, utilisation and maintenance costs?
Limits of application
The figures are an idealised hypothetical example, not an estimate of Google’s or any other company’s satellite design or actual area. σ is the Stefan–Boltzmann constant; steady state, uniform temperature and one radiating face are assumed. Actual design requires incoming radiation, heat transport, multiple surfaces, material degradation and margins, and this calculation alone cannot establish commercial viability or share prices.
- What to look for
- Follow how the investment argument about AI bottlenecks leads to technical and economic assumptions about orbital data centres and long-term portfolio proposals.
- Why it matters
- Separate a successful test from scaling economics and check heat rejection, launch, maintenance and utilisation alongside power. This helps test the leap from technological need to a particular stock’s returns.
Timeline
7 segmentsGoogle’s announcement confirms orbital insertion and contact with a Suncatcher prototype, but describes a research mission collecting data on radiation and thermal conditions. NASA thermal-control material explains that a vacuum has no external air convection, so a cold space environment does not eliminate cooling hardware. Large-scale deployment and chip-production plans require evidence on lifetime, performance, demand and total cost; the video’s company timelines and revenue links do not establish commercial viability.