Green Distributed AI Training: Orchestrating Compute Across Renewable-Powered Micro Datacenters

2025-11-20

AI Summary

To address the spatiotemporal mismatch between intermittent renewable energy and centralized data centers, this paper proposes a dynamic green AI training architecture based on a network of micro-datacenters. Methodologically, it introduces a “feasibility domain” model that quantifies temporal constraints as dominant factors in workload migration, thereby formulating migration as a tractable, controllable time-series optimization problem. The architecture integrates checkpoint-driven migration, wide-area bandwidth-aware scheduling, surplus renewable energy window forecasting, and grid-coordinated orchestration, while leveraging edge-optimized GPU platforms. Experiments demonstrate that the approach reduces fossil fuel dependency by up to 47%, enhances training stability and energy efficiency, and supports both partially migratable and distributed AI workloads. It achieves a balanced trade-off between performance and sustainability in energy-compute co-optimization.

Technology Category

• Machine Learning: Efficient ML / Green AI
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Sustainability and carbon-aware systems for Web, mobile, and WoT
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work

Abstract

The accelerating expansion of AI workloads is colliding with an energy landscape increasingly dominated by intermittent renewable generation. While vast quantities of zero-carbon energy are routinely curtailed, today's centralized datacenter architectures remain poorly matched to this reality in both energy proportionality and geographic flexibility. This work envisions a shift toward a distributed fabric of renewable-powered micro-datacenters that dynamically follow the availability of surplus green energy through live workload migration. At the core of this vision lies a formal feasibility-domain model that delineates when migratory AI computation is practically achievable. By explicitly linking checkpoint size, wide-area bandwidth, and renewable-window duration, the model reveals that migration is almost always energetically justified, and that time-not energy-is the dominant constraint shaping feasibility. This insight enables the design of a feasibility-aware orchestration framework that transforms migration from a best-effort heuristic into a principled control mechanism. Trace-driven evaluation shows that such orchestration can simultaneously reduce non-renewable energy use and improve performance stability, overcoming the tradeoffs of purely energy-driven strategies. Beyond the immediate feasibility analysis, the extended version explores the architectural horizon of renewable-aware AI infrastructures. It examines the role of emerging ultra-efficient GPU-enabled edge platforms, anticipates integration with grid-level control and demand-response ecosystems, and outlines paths toward supporting partially migratable and distributed workloads. The work positions feasibility-aware migration as a foundational building block for a future computing paradigm in which AI execution becomes fluid, geographically adaptive, and aligned with renewable energy availability.

Problem

• Developing distributed AI training across renewable-powered micro datacenters
• Addressing mismatch between intermittent renewable energy and centralized datacenters
• Enabling workload migration to follow surplus green energy availability

Innovation

• Distributed micro-datacenters dynamically follow surplus green energy
• Feasibility-domain model links checkpoint size, bandwidth, and renewable duration
• Orchestration framework transforms migration into principled control mechanism