Emerald and Lumen explore power-aware routing
A new idea for moving AI requests
Emerald AI and Lumen announced a collaboration on October 29 to route AI inference requests among networked locations based on power availability, network conditions and cost. Inference is the work of using a trained model to answer a request; it differs from training the model. The companies say the goal is to steer requests toward places with available electricity.
Emerald’s Conductor software would provide power information; Lumen’s programmable fiber network would connect locations and help route traffic. The firms are exploring a pilot using Lumen’s edge and network infrastructure, but they name no site, customer, start date or operating test.
The network becomes part of the power plan
A request can move only if a remote location has the required model, computing capacity and secure data access. The network must then return an answer quickly enough for the application. The companies say their design would account for latency, security, network conditions and cost; these are requirements, not published performance results.
Analogy: It resembles sending an order to another branch when one store has limited stock. That helps only if the second branch has what the customer needs and the extra trip is still fast. The comparison explains the routing idea; it is not a measured data-center result.
Possible flexibility, not a demonstrated grid service
If the pilot works as intended, shifting some flexible inference requests could reduce demand at one location during a local power constraint while another site handles the work. Neither company reports electricity reductions, cost savings or an effect on household rates. The release describes a planned test, not an operating service.
Evidence to look for
A useful pilot report would identify its sites, workloads, response times, security safeguards and before-and-after power readings. Those details would show whether network routing can ease a local constraint without making service slower or less reliable.
Understand the infrastructure
Read the related Grid & Silicon explainer →
Sources linked below.
Continue your research
Read the original source at Lumen Technologies announcement ↗
Explore the freely accessible source for additional detail and context. This is a Grid & Silicon summary, not firsthand reporting.