The team developed FABRIC, short for Fabrication-to-Grave Biodiversity Impact Calculator, the first modeling framework to trace computing’s biodiversity footprint across its lifecycle: manufacturing, transportation, operation, and disposal.

Computers power everything from artificial intelligence to climate modeling, but a new study from Purdue University’s Elmore Family School of Electrical and Computer Engineering shows that this progress may come with an overlooked environmental cost: biodiversity loss.

In a first-of-its-kind paper presented at the 4th Workshop on Sustainable Computer Systems (HotCarbon’25), Assistant Professor Yi Ding and ECE PhD student Tianyao Shi unveiled the first framework to measure how computing systems, from chip fabrication to data center operation, affect global ecosystems and species diversity.

“Traditionally, sustainability research in computing has focused on carbon emissions and water consumption,” Ding explained. “But biosphere integrity is a critical planetary boundary. Our work provides the first quantifiable link between computing activities and biodiversity loss.”

FABRIC: Connecting Computing to Ecology

The team developed FABRIC, short for Fabrication-to-Grave Biodiversity Impact Calculator, the first modeling framework to trace computing’s biodiversity footprint across its lifecycle: manufacturing, transportation, operation, and disposal.

To make these impacts measurable, Ding’s group introduced two new metrics:

• Embodied Biodiversity Index (EBI): Captures the one-time environmental toll of manufacturing, shipping, and disposing of computing hardware such as CPUs, GPUs, and memory.

• Operational Biodiversity Index (OBI): Measures the ongoing biodiversity impact from the electricity used to power computing systems.

The analysis integrates data on pollutants like sulfur dioxide, nitrogen oxides, and heavy metals—key drivers of acid rain, eutrophication, and freshwater toxicity—translating them into a unified “species·year” metric that represents the fraction of species lost in an ecosystem over time.

What They Found

Using FABRIC, the team analyzed seven high-performance computing (HPC) workloads running on a range of hardware, from local servers to supercomputers and cloud platforms. Their findings were striking:

• Manufacturing dominates the embodied impact, responsible for up to 75% of total biodiversity damage, largely due to acidification from chip fabrication.
• Operational electricity use overshadows manufacturing—at typical data center loads, the biodiversity damage from power generation can be nearly 100 times greater than that from device production.
• Newer, more efficient devices are better for biodiversity per unit of performance, but their total footprint still depends heavily on energy sources.
• Location matters: Renewable-heavy grids with strict sulfur and nitrogen emission limits—like Québec’s hydroelectric mix—can cut biodiversity impact by an order of magnitude compared to fossil-fuel-heavy grids.

“People often equate sustainability with reducing carbon,” Ding said. “But our study shows that low carbon doesn’t always mean low biodiversity impact. A coal-heavy grid might have similar carbon emissions to a gas-heavy one but much higher acid gas emissions that harm ecosystems.”

From Energy Efficiency to Ecological Efficiency

Ding’s interest in sustainability stems from her doctoral work in energy-efficient computing, which she has since expanded to include the broader environmental effects of technology.

“I realized that energy efficiency is just one part of the sustainability puzzle,” she said. “We also need to understand how computing affects the planet’s living systems.”

This interdisciplinary work—conducted in collaboration with Inez Hua, a Purdue faculty in School of Sustainability Engineering and Environmental Engineering, and Lyles School of Civil and Construction Engineering—demonstrates the value of cross-campus research. “Purdue fosters an environment where engineers and environmental scientists can work together,” Ding added. “That collaboration made this project possible.”

A New Frontier for Sustainable Computing

As data centers grow and AI workloads surge, Ding believes biodiversity must become a “first-class metric” in system design.

“Our goal isn’t to stop progress—it’s to make computing more aware of its ecological footprint,” she said.

By introducing a method to quantify biodiversity impact, Ding’s work adds a new dimension to the sustainability conversation in technology—one that connects servers to species, and algorithms to ecosystems.

The team’s paper, “When Servers Meet Species: A Fab-to-Grave Lens on Computing’s Biodiversity Impact,” is available on arXiv.

About the Research Group

The Sustainable Computing and Infrastructure Lab led by Yi Ding focuses on building environmentally responsible computing systems by integrating sustainability metrics, carbon, water, and biodiversity, into the design, operation, and optimization of large-scale computing infrastructure.