200GW Hiding in Grid, Sodium Batteries 10x Cheaper, Wave-Powered Datacenters w/ Ramez Naam | EP #280 — Key Takeaways
YouTube
200GW Hiding in Grid, Sodium Batteries 10x Cheaper, Wave-Powered Datacenters w/ Ramez Naam | EP #280
Peter H. Diamandis2h 6mAug 15, 2026
Watch the originalGetting power to a new U.S. data center before 2031-2032 is nearly impossible today — the grid's poles and wires, not generation capacity or cost, are the binding constraint on AI growth.
Key takeaways
Grid interconnection wait times doubled in 4 years — not cost, not generation
Grid interconnection wait times doubled in 4 years — not cost, not generation
- Connection queue for new power projects grew from 15 months (20 years ago) to 45 months today.
- Texas, the least-regulated US grid, won't fulfill a new data center power request before 2031-2032.
Being an 'interruptible load' cuts grid connection time from 7 years to 12-18 months
Being an 'interruptible load' cuts grid connection time from 7 years to 12-18 months
- Texas June 2024 regulation: data centers that accept grid operator shutoff during peak get fast-tracked.
- FERC then sent letters to the six other largest US grids telling them to adopt similar rules.
Fusion is regulated like hospital imaging machines, not fission reactors — a huge industry unlock
Fusion is regulated like hospital imaging machines, not fission reactors — a huge industry unlock
- This regulatory decision, made quietly during COVID, means fusion projects avoid the multi-decade NRC licensing burden of fission plants.
- The key distinction: fission can melt down if coolant fails; fusion simply stops — it is default-off, not default-on.
This Dig holds 4 more insights, 4 flashcards, and 3 quotes — free with your trial.
Unlock this Dig freeStart free with 100 credits · No card, no expiry
In this video
- 1mIntroduction to the energy landscape and technological potential
- 2mGuest introduction: Remez and his work on energy futures
- 5mAI's power hunger and the need for grid expansion
- 9mLog-linear relationship between compute and AI progress
- 15mMoving data centers to where energy is abundant
- 18mThe potential of large-scale solar projects and their costs
- 21mThe role of nuclear fission and fusion in future energy
- 34mThe analogy of cache and multitasking in energy management
- 40mThe scale of current and future solar installations
- 45mThe economics of GPUs, AI, and energy costs
- 50mLong-term prospects for solar, nuclear, and space-based energy
“10 gawatt a year is like five or six launches of Starship a day.”
— Romez Naam
This page is a partial, transformative summary produced by Homestake. All rights to the original content remain with its creator — please support them at the source link above.
