PROMETHEUS WP-03
COMPUTE FOLLOWS ENERGY
The Geography of Intelligence
Artificial intelligence may be digital, but computation is constrained by geography. WP-03 examines how the location of energy, computing infrastructure, communications networks, and data can shape where intelligence is produced—and how suitable computational workloads may increasingly move toward available energy rather than requiring all energy to move toward compute.
What WP-03 Examines
The Geography of Intelligence
Intelligence infrastructure has a geography. Large-scale computing depends upon access to electricity, fiber connectivity, data centers, cooling, land, equipment, and other physical resources. WP-03 examines how these constraints can influence where computation is located—and why the geography of available energy may increasingly shape the geography of artificial intelligence.
The Virtual Transmission Line
Electricity does not always have to move to computation. In some cases, suitable computational workloads can move toward available electricity through high-capacity communications networks. WP-03 describes this relationship as a Virtual Transmission Line: instead of physically transmitting every additional unit of power across the grid, computation can be shifted between connected regions when workload characteristics, latency, data requirements, and infrastructure allow.
The Continental AI Campus
The emerging geography of intelligence may increasingly resemble a continental-scale computing system rather than a collection of isolated data centers. WP-03 introduces the Continental AI Campus: a network of geographically distributed compute regions connected by high-capacity communications infrastructure, allowing suitable workloads to operate where combinations of energy, computing capacity, connectivity, and other physical resources are available.
The Compute-Energy Operating System
Coordinating geographically distributed intelligence infrastructure requires more than connecting data centers with fiber. WP-03 introduces the concept of a Compute-Energy Operating System: a framework for coordinating computational workloads with available energy, network capacity, latency requirements, data location, and computing resources across multiple regions. The objective is not to move every workload, but to identify which forms of computation can move—and when doing so improves the use of both energy and compute infrastructure.
The Limits of Moving Compute
Not every computational workload can move freely. Latency, data gravity, network capacity, specialized hardware, security requirements, and the characteristics of individual workloads can limit where computation can occur. WP-03 therefore treats geographic flexibility as a bounded capability: some forms of compute may move toward available energy, while others must remain close to users, data, equipment, or other critical infrastructure.
Building the Geography of Intelligence
WP-03 ultimately asks how intelligence infrastructure can be designed when energy, compute, communications, and geography are treated as parts of the same system. The emerging architecture may combine regional computing centers, high-capacity networks, geographically distributed energy resources, and workload flexibility into a larger infrastructure capable of directing suitable computation toward the places where the physical resources required to support it are available.
Read the WP-03 Gold Edition
PROMETHEUS WP-03: Compute Follows Energy examines how the geography of energy may increasingly shape the geography of artificial intelligence. From the Virtual Transmission Line and Continental AI Campus to the Compute-Energy Operating System, WP-03 explores how energy, computation, communications networks, and geography can be coordinated as components of an emerging infrastructure for intelligence.