Salesforce: A Green Plan for AI's Data Centre Impact

Salesforce, a well-known player in AI and data solutions, has published its AI Sustainability Outlook report, bringing attention to the pressing energy demands that AI technologies are placing on the world's data centre infrastructure.

The report emphasises the necessity for redesigning these centres with sustainability at the forefront to curb energy consumption.

“As a leader in agentic AI, it is Salesforce's imperative to ensure that AI is trusted, reliable and sustainable," Sunya Norman, SVP Impact at Salesforce, says in a post on LinkedIn.

“In this outlook, we explore the current landscape, our efforts and evolving insights so far and our preliminary path forward.

“We’re early in the journey, but the future is being shaped now.

“By sharing our progress openly, we aim to spark transparency and inspire collective action.

“A sustainable future with AI is within reach and I’m optimistic about what we can achieve together.”

The environmental cost of AI

The AI Sustainability Outlook from Salesforce predicts an alarming increase in global electricity consumption by AI-related data centres, estimating they could consume about 3% of the global supply by 2030.

This surge is a result of the compute power required to manage AI processes, coupled with sophisticated cooling systems in place to maintain optimal environment conditions for running advanced algorithms.

Further compounding this challenge, Gartner analyses suggest that up to 40% of AI data centre projects could experience significant power bottlenecks by 2027. These bottlenecks could directly contribute to delays, escalated costs and issues with power reliability, showcasing a crucial need for system upgrades.

Right now, about 56% of energy powering data centres is derived from non-renewable fossil fuels.

If this reliance continues unchecked, the International Energy Agency (IEA) warns that data centres could become one of the most rapidly increasing sources of carbon emissions worldwide.

Additionally, AI operations demand substantial water usage for cooling — one-fifth of which stems from water-stressed regions.

The construction of AI compute infrastructure also necessitates large quantities of minerals like lithium and copper, further taxing environmental resources and contributing to broader ecological degradation.

AI as a sustainability tool

While AI's growth in the data centre industry presents environmental challenges, Salesforce underscores its potential in accelerating sustainability efforts.

According to the IEA, AI innovation has the capability of reducing energy-related emissions by 5% by 2035.

AI technologies can optimise systems for predictive maintenance, streamline renewable energy integration, deliver customised energy education and refine climate prediction models.

In water management, AI offers solutions for intelligent irrigation systems, enhances wastewater treatment operations and proficiently monitors water quality.

Salesforce’s proprietary AI platform, Agentforce, exemplifies the practical application of AI in sustainable projects.

A notable instance is Good360's collaboration, where its AI resource matching system is revolutionising the distribution of aid in disaster scenarios, effectively reducing waste and promoting sustainable practices.

Stephane Moulec, Chief Technology Officer at Good360, says: “Globally, a significant amount of goods that could be matched to disaster survivors end up going to the landfill.

“Good360 is here to change that.”

Rare is developing an AI-powered regenerative agriculture coach, delivering real-time guidance to farmers using local crop, weather and environmental data, aiming to cut staff time by 40%.

Groundswell is leveraging AI to expand community solar access, helping 30,000 households reduce energy costs by half.

Salesforce’s three-pillar strategy

Salesforce champions a strategic approach built on three key pillars to mitigate AI’s environmental consequences: smart demand, efficiency and clean supply.

Within the smart demand framework, Salesforce advises the precise selection of AI tools tailored to specific project requirements.

This ensures that resources are not overextended by energy-intensive AI models that are not necessary.

The company also emphasises the importance of transparency with clients regarding power consumption data, incentivising solutions that improve operational efficiency.

Salesforce advocates for efficiency by recommending the deployment of smaller, specialised AI models that bypass the exhaustive computational demands of general-purpose alternatives.

Techniques like model quantisation, distillation and pruning are promoted to reduce compute needs, permitting smaller models to operate efficiently on smaller hardware rather than massive data centres.

For the clean energy supply pillar, Salesforce encourages the integration of sustainability into procurement activities.

This involves partnering with environmentally-conscious suppliers while investing in renewable energy projects and sustainable water resources.

Additionally, supporting policy and infrastructure changes at an aggregate level is crucial for broad-scale environmental impact mitigation.

By coherently aligning AI deployment with sustainable methodologies, Salesforce anticipates the data centre sector can seamlessly accommodate AI’s rapid advancement while curtailing its ecological imprint.