Salesforce Introduces AI Energy Score: A Game-Changer for Sustainable AI Development
In a major leap toward creating more sustainable artificial intelligence, Salesforce, in collaboration with Hugging Face, Cohere, and Carnegie Mellon University, has unveiled the AI Energy Score. Launched at the AI Action Summit on 10 February 2025, this pioneering benchmarking tool aims to provide transparency around the environmental impact of AI models, offering a clear and standardised way to measure their energy efficiency.
The AI Energy Score is designed to tackle a critical issue in the AI sector: the energy consumption of AI models and their effect on the environment. As artificial intelligence continues to grow and influence industries worldwide, the need to understand and mitigate its ecological footprint has become increasingly urgent. Salesforce’s initiative offers a transparent framework for developers and users to assess and compare the energy use of AI models, making sustainability an integral part of AI adoption.
This benchmarking system introduces a range of features, including standardised energy ratings, a public leaderboard for 166 commonly-used AI models, and a recognisable 1-to-5 star energy label. The aim is to drive market preference towards more energy-efficient models and encourage the adoption of sustainable AI practices across industries.
“The AI Energy Score is a first-of-its-kind tool that is set to become a critical enabler of responsible AI development,” said Boris Gamazaychikov, Head of AI Sustainability at Salesforce. “We are at a critical juncture with the rapid acceleration of both the climate crisis and AI innovation. This score offers a transparent and standardised method to assess AI model energy consumption, helping to reduce operational costs, optimise infrastructure, and ensure long-term sustainability.”
The initiative has already gained traction, with the French Government and the Paris Peace Forum recognising the AI Energy Score’s potential to transform the AI ecosystem. It is expected to foster greater accountability, helping developers and organisations make informed decisions about their models’ efficiency and carbon footprint.
Setting New Standards in AI Energy Efficiency
The AI Energy Score stands out as a comprehensive tool for AI developers. It features a public leaderboard that ranks AI models based on their energy efficiency across ten common tasks, such as text generation, image generation, and summarisation. By showcasing the energy scores of well-known models, such as Salesforce’s SFR-Embedding, xLAM, and SF-TextBase, the leaderboard helps users quickly identify which models are the most sustainable.
Additionally, the AI Energy Score offers a benchmarking portal where developers can submit both open-source and proprietary models to be evaluated. Open models can be automatically tested, while closed models can be evaluated through a secured testing sandbox, ensuring fairness and consistency in energy assessments.
The introduction of a recognisable energy label, from 1 to 5 stars, makes it easy for developers and users to identify energy-efficient models. This simple labelling system is expected to drive demand for low-energy AI solutions, giving companies a competitive advantage and contributing to a greener AI ecosystem.
Salesforce’s Commitment to Sustainable AI
Salesforce has long been a champion of innovation that prioritises sustainability, and its Agentforce platform exemplifies this approach. Launched in 2024, Agentforce empowers businesses to deploy autonomous AI agents across various functions while focusing on energy efficiency. Unlike traditional AI models that require resource-intensive training, Agentforce is designed to optimise energy consumption right out of the box.
“Agentforce combines efficient small language models with agentic reasoning, reducing energy consumption without sacrificing performance,” explained Suzanne DiBianca, EVP and Chief Impact Officer at Salesforce. “By minimising computational waste, Agentforce delivers high accuracy and responsiveness while lowering the environmental impact of AI operations.”
The launch of the AI Energy Score marks another step in Salesforce’s journey towards responsible AI development. The company will be the first AI model developer to disclose the energy efficiency data of its proprietary models under this new framework, setting an example for others to follow.
“We are proud to lead by example,” DiBianca added. “By adopting practices that reduce energy consumption, we can help businesses lower their operational costs, optimise their infrastructure, and improve their long-term sustainability.”
A Collaborative Effort for a Sustainable Future
The AI Energy Score is the result of a collaborative effort involving Salesforce, Hugging Face, Cohere, and Carnegie Mellon University. By establishing a standard for energy consumption in AI, the initiative aims to drive widespread change and make sustainable AI development the norm rather than the exception.
Dr. Sasha Luccioni, AI & Climate Lead at Hugging Face, emphasised the importance of this transparency. “By creating a transparent rating system, we address a key blocker to reducing the environmental impact of AI. This initiative sets an important milestone for sustainable AI.”
The launch of the AI Energy Score also comes at a critical time. Recent surveys indicate that 65% of sustainability professionals recognise the need to balance the benefits of AI with environmental concerns, particularly as the energy demands of AI-driven data centres are expected to double by 2026.
As AI adoption accelerates, the AI Energy Score offers a vital tool for ensuring that AI models are both effective and environmentally responsible. By helping businesses assess and optimise their energy use, it promotes a more sustainable future for the AI industry and the planet as a whole.
A Call for Industry-Wide Change
Salesforce’s commitment to sustainable AI through the AI Energy Score reflects a broader industry trend towards accountability and transparency. By setting a new standard for AI energy efficiency, this initiative has the potential to revolutionise how AI is developed, deployed, and evaluated across sectors.
“The AI Energy Score is more than just an energy-use metric,” said Gamazaychikov. “It’s a strategic business advantage that establishes a clear, trusted benchmark for sustainability, enabling companies to make informed decisions and embrace practices that reduce their carbon footprint.”
As AI continues to evolve, the AI Energy Score promises to play a central role in shaping a sustainable and responsible future for the technology, making it possible to harness the power of AI while protecting the planet.
Source: https://www.salesforce.com/uk/news/stories/ai-energy-score/
