๐Ÿ”ฎ Sneak Peak Saturday for 2025-09-06

Sometimes, $600B just comes out in polite conversation. Twice?

Of course, one of the $600B by 2028 is hard to fathom considering the most recent number for 2025 is only $65B per Meta.

What could $600B mean? Well, what if it applied to:

  • Selection of strategic global locations
    • US Southwest, Northeast, and Central
    • Sweden
    • Australia
  • Commitment to 3 year "get to greenโ€™ AI data center designs
    • strategic land purchases for thousands of acres
    • state and regional vPPA negotiations with investments
  • Construction of data centers optimized for AI workloads
    • procurement of best-in-class AI infrastructure
    • construction of solar power generation sites
    • construction of power-storage facilities
    • commissioning SMR power infrastructure designs
  • Consumption of AI infrastructure
    • procurement of best-in-class AI infrastructure
    • evolving with the latest generation of AI chips
    • begin transition to custom silicon AI chips
    • begin addition of quantum computing
  • Give excess power back to the grid by 2030
    • convert to 100% solar and SMR
    • fully transition to custom silicon AI chips
    • fully transition to custom silicon and quantum computing

What If

This would be the equivalent of a cookie-cutter approach to building multiple sites with a Topaz Solar Farm, dual SMRs, and enough vPPA to keep an AI-first design data center footprint humming along. Also, by adding to and evolving the latest generation of AI chips with an eye towards the quantum computing future, it could easily use up $600B (or more) in less than 3 years.

Oh, and just by setting aside even ~$30B might enable some of that AI compute to be flying over our heads one day.


View this page on GitHub.


NexusTekThanks to NexusTek for sponsoring the free version of this newsletter post ๐Ÿค“