The AI Boom Has a Water Problem. Salesforce Is Making It a Priority.

Salesforce has been on a sustainability journey for more than a decade and established sustainability as one of the company’s five core values. As part of this commitment, Salesforce published its Nature Positive Strategy in 2023 to increase transparency and guide action on its impacts and dependencies on nature. While carbon reduction remains the company’s top priority across its full environmental footprint, the strategy identified water as a significant nature-related priority – essential to both ecosystem health and business resilience.

In this age of rapidly expanding use of AI, concerns about the industry’s effects on the natural world –particularly those related to energy–have garnered attention because AI systems can be computationally intensive. They come with energy demands that could strain our aging electric grid. Meeting those energy needs with fossil fuels could give the tech industry a larger carbon footprint.

The increased use of clean, renewable energy to power the AI revolution is a key part of the company’s Nature Positive Strategy. However, Salesforce’s own deep dive into its impacts on the Earth found water to be an underappreciated factor – and one that is deeply interconnected with energy.

Now, the company is taking new steps to restore waterways and funding new programs to boost water conservation worldwide. “We're updating our Nature Positive Strategy with a dedicated water program, because water is our biggest material footprint,” said Tim Christophersen, Vice President for Climate Action at Salesforce.

Building Water Resilience

As the world’s leading customer relationship management (CRM) technology, Salesforce has always been in the business of building and improving other businesses’ customer relationships. Nowadays those relationships increasingly occur via AI systems, like Agentforce that can field customers' questions and take action to help solve problems.

Those AI systems can require the computational power of a growing armada of data centers around the world, many of which have a water footprint.That is largely because computing systems need to be cooled. Although some new facilities are equipped with more sophisticated cooling systems, most existing data centers rely on water for cooling, meaning they can pull in and then discard large quantities from bodies of water or water systems situated near the facility.

This water usage can potentially disrupt the ecosystems around the facility. When looking at water impacts in Salesforce’s value chain, Christophersen found that these computational powerhouses account for 80% of its water usage – and that the numbers for the industry at large are considerable. “If this continues unchecked with the current AI boom,” he said, “then AI-triggered data center usage could use as much water as the country of Denmark does by 2030.”

Salesforce is expanding its water focus with new investments to protect critical waterways across Latin America. In Brazil, the company is supporting Conservation International’s work in the Jaguari River Basin — a key water source for nearly 9 million people — restoring springs, riparian zones, and native vegetation to strengthen watershed resilience.

In Mexico, Salesforce is backing three watershed initiatives: restoring the Xochimilco wetland, reforesting the Cutzamala and Moctezuma watersheds that supply Mexico City, and supporting Forests for Monarchs in Michoacán to restore monarch habitat and advance sustainable land management with Indigenous and local communities.

“Based on our water and nature assessment, we have identified a number of priority locations around the world where our investments in reforestation and forest ecosystem restoration can improve the water absorption capacity, the water filtration capacity of those ecosystems, and the capacity to replenish water sources,” Christophersen says.

These commitments are part of a three-pronged approach to protecting this precious resource.

• Investing directly in conservation efforts on crucial waterways to make these ecosystems more resilient.
• Pushing for a more resilient, renewable power supply for data centers to make a region’s energy industry less water-intensive and reduce the pressure on those waterways.
• Engaging suppliers to consider infrastructure changes to data centers, like water conservation or reclamation.

Although water conservation is the most recent focus of the Nature Positive Strategy, Salesforce continues to take action on its ambitious long-term nature goals. Most notably Salesforce is helping to fund the planting of 100 million trees as part of 1t.org, of which Salesforce is a founding partner. The company is also contracting to purchase 1 million tons of blue carbon (C02 captured and stored in marine ecosystems to keep the greenhouse gas out of the atmosphere) and it created a $100 million nature restoration fund. But the company cannot go it alone, Christophersen says. “By building on relationships across various sectors of the economy, Salesforce and its partners can begin to envision an AI-driven future that is a sustainable one, too.”

On a Journey Towards More Sustainable AI

"The immediate challenge is developing a vision for 'more sustainable AI' – one focused on measurable reductions," Christophersen said. With the industry so nascent and growing so quickly, industry leaders are figuring out as they go how to reduce the water, carbon, and other natural footprints of AI systems. For Salesforce, sustainable AI includes all the ways the company and its customers, partners, and communities are working to reduce the planetary impacts of this growing sector. Salesforce is focused on three pillars of AI sustainability: smart demand, efficiency, and clean supply.

• Clean Supply: Salesforce engages its AI supply chain, from cloud providers to chip manufacturers, to strengthen renewable energy sourcing, water stewardship, and sustainable procurement practices.
• Efficiency: Using model quantization and code pruning to cut energy use and optimize data center performance. Running models on Salesforce’s Hyperforce infrastructure further boosts efficiency – delivering up to 40% lower energy consumption than traditional hosting environments.
• Smart Demand: Choosing right-sized, domain-specific models instead of defaulting to large, general LLMs. Agentforce 360, Salesforce’s AI platform, is architected to reuse compute across tasks, reducing redundant training while pricing models align usage with both cost and carbon impact.

“This is where Salesforce can help: by identifying what is actually the minimum amount of data that is of the highest quality for use with an LLM that is specific to a company or to a company's main use,” he said. “If you use large, generally trained LLMs for small tasks that are very specific and internal to a business's specific needs, you waste a lot of energy.”

Restoring our Relationship with Nature

Climate change, pollution, and biodiversity loss are symptoms of a broader crisis related to the way we think about the Earth and our place on it, leading to this unwinnable war against nature we now find ourselves in. “The relationship that we currently have is only exploitative and extractive,” he said. “How can we build a different relationship that is more reciprocal?”

With the Nature Positive Strategy and its new water-focused phase, Salesforce is pursuing a vision of sustainable planning and collaboration with nature where an AI-powered future can exist without destroying ecosystems and over-consuming resources.

“All of that starts with learning from nature,” he said. “It's reconciling ecology and economy, which have the same root in ancient Greek. If we manage something without knowing how it works, then you see what is happening right now – and that is something we need to fix.”