Science Friday

Secrets Of Ancient Concrete, And... Data Centers In Space?

January 14, 2026

Key Takeaways Copied to clipboard!

  • The discovery of an active construction site in Pompeii revealed that ancient Romans likely used a "hot mixing" technique for concrete—mixing quicklime and volcanic ash dry before adding water—which contributes to its remarkable self-healing properties that modern concrete lacks. 
  • Tech leaders are considering moving power-hungry data centers to space to harness continuous solar energy, but face significant engineering hurdles related to launch costs, massive solar panel surface area, radiative cooling requirements, and data transfer limitations. 
  • While launching massive terrestrial data centers to space is a long-term ambition, near-term opportunities exist for smaller-scale orbital computing, such as processing data collected by satellites before sending it back to Earth. 

Segments

Ancient Concrete Discovery Site
Copied to clipboard!
(00:00:27)
  • Key Takeaway: An active construction site preserved in Pompeii’s volcanic ash provided direct evidence of Roman concrete mixing methods.
  • Summary: Host Ira Flatow expressed his long-standing interest in concrete durability, citing Roman examples like the Pantheon and aqueducts. Scientists discovered a construction site in Pompeii frozen in time by the eruption, offering a snapshot of raw materials, tools, and mixing processes. This site allowed researchers to directly observe how Roman concrete was prepared.
Roman Concrete Recipe Revealed
Copied to clipboard!
(00:03:04)
  • Key Takeaway: Ancient scholars like Vitruvius documented mixing volcanic ash and lime, but the Pompeii site suggests Romans mixed quicklime and ash dry before adding water, a process called ‘hot mixing.’
  • Summary: Ancient texts from Vitruvius and Pliny the Elder documented recipes for Roman cement, noting its ability to set underwater. Vitruvius suggested slaking quicklime (calcium oxide) with water first before mixing it with ash. However, the Pompeii evidence shows the calcined stone (quicklime) was ground, mixed dry with volcanic ash, and then water was added, causing the mix to heat up significantly.
Self-Healing Concrete Mechanism
Copied to clipboard!
(00:05:35)
  • Key Takeaway: The Roman ‘hot mixing’ technique results in a self-healing concrete where microcracks dissolve and recrystallize, a property absent in modern concrete analogs.
  • Summary: The self-healing quality of ancient Roman concrete is attributed to the hot mixing process. When microcracks form, the reaction allows for dissolution and recrystallization within those cracks. Dr. Masic confirmed that his team has patented ideas to incorporate this Roman-inspired self-healing mechanism into modern concrete.
Challenges of Space Data Centers
Copied to clipboard!
(00:08:29)
  • Key Takeaway: Moving data centers to space is driven by the need for massive, continuous solar power, but faces severe challenges in launch costs, cooling, and data transfer bandwidth.
  • Summary: The primary motivation for orbital data centers is securing immense, uninterrupted power, as terrestrial grids struggle to supply the necessary gigawatt loads, and Earth-based solar is intermittent. Cooling in space requires relying on radiative cooling, necessitating large surface area panels separate from the solar collection panels. Furthermore, transferring massive datasets (like those needed for AI training) back to Earth is difficult, potentially requiring launching the data disks physically.
Efficiency vs. Orbital Ambition
Copied to clipboard!
(00:14:38)
  • Key Takeaway: Current AI development prioritizes generality over efficiency, but specialized models and future energy infrastructure improvements on Earth may negate the immediate need for large-scale orbital data centers.
  • Summary: The current pursuit of highly general AI models consumes vast energy resources without prioritizing efficiency. Developing specialized models for specific domains (finance, medicine) could drastically reduce computational costs. In the next 10-15 years, improvements in terrestrial energy capacity (batteries, solar, SMRs) might alter the cost-benefit analysis of sending infrastructure to space.