Google has introduced a new approach to managing its data centre infrastructure that aims to significantly reduce emissions by treating computing resources as a shared pool rather than discrete, team-owned machines.
The programme, called Central Fleet, allows internal teams to request computing capacity—processing power, memory, storage—via an “intent-based” system rather than ordering specific hardware. These requests are fulfilled from a centralised, fungible pool of resources, meaning Google can reuse existing infrastructure rather than procuring new machines for each project.
Key benefits highlighted by Google include:
- Reduced embodied carbon: In 2024, Central Fleet avoided procurement of new machines and components, saving an estimated 260,000 metric tons of CO₂ equivalent.
- Less electronic waste thanks to reallocation and reuse of under-utilised hardware.
- Greater energy efficiency, by placing workloads on the most power-efficient or appropriate hardware available.
- Supporting a circular economy model rather than the traditional linear take-make-dispose pattern.
The infrastructure enabling this includes Google’s Borg cluster management system, which abstracts the physical machines into a software-defined, flexible resource layer. This allows resources to be reallocated dynamically, previous resources to be returned to the pool, and reduces the need for over-provisioning.
By integrating sustainability into its base infrastructure operations, Google aims to meet growing demand for cloud and AI services while lowering its environmental impact.
