AI adoption in sustainability has surged in the past year. Companies use AI to cut energy use, reduce waste, enhance workplace safety, and accelerate toward their sustainability goals. Early results are promising. Leaders are uncovering emissions hot spots in real time, optimizing renewables deployment, and transforming employee experience.

This is just the beginning. Of the 400 C-suite and sustainability executives recently surveyed by Bain & Company in eight markets, almost 80% say they see high or very high opportunity for AI to contribute to their sustainability agendas. Yet more than 50% are still in the initial stages of piloting and exploring these AI applications—underscoring how much potential remains to scale AI that accelerates real-world progress on sustainability.

This is good news, but scaling AI will also bring true environmental and workforce costs that must be carefully managed. Our proprietary climate-economic modeling tool, Intersect^SM, finds that, in a high-growth scenario, AI and data centers could emit 810 million metric tons of carbon dioxide annually by 2035, or 2% of global emissions and 17% of industrial emissions. The carbon footprint of AI is closely tied to the source of electricity powering data centers. In regions such as the US, where a higher portion of the grid is still dependent on fossil fuels, the share of AI-driven industrial emissions could increase from 18% in 2022 to more than 50% by 2035. By contrast, Europe’s accelerated transition to renewable energy and more measured AI adoption are expected to keep emissions relatively stable (see Figure 1). While hyperscalers and large data center operators have committed to renewable energy, the urgency of deployment and current build-out constraints mean emissions from AI will likely reflect the regional electricity mix in the near term.

To put this in perspective, in this scenario, the emissions added between 2025 and 2035 would cancel out three years’ worth of emissions cuts by the world’s 500 largest disclosing companies.

While this setback may diminish as electric grids decarbonize, AI’s impact on employment remains far more uncertain and difficult to prepare for. In the AI era, revenue is increasingly decoupled from workforce growth. Average revenue per employee in real terms rose from $840 between 2016 and 2020 to $1,090 between 2021 and 2024, and that gap is widening. Output is rising without equivalent job creation, reshaping the future of work faster than most organizations are prepared to handle. This is no longer a distant concern; CEOs openly recognize that AI is transforming their workforces, particularly white-collar jobs.

While many companies are still grappling with AI’s trade-offs, a small but powerful group is showing what’s possible. We identified these leaders—whom we call “shapers” and who represent about 20% of the companies in our study—based on four criteria: their maturity in scaling AI for sustainability, the breadth of sustainable AI use cases adopted, the value they report from using AI to advance sustainability, and their overall maturity in sustainability.

The key differentiator isn’t who leads in sustainability; it’s who leads in AI. Shapers are deeply committed to AI’s potential to tackle complex sustainability challenges. They use AI in sustainability efforts three times more often than laggards and are nearly two times more likely to focus on its long-term value creation (see Figure 2). Although they’re found across industries and regions, shapers are most prevalent in the technology and manufacturing sectors.

Figure 2

Shapers’ strategies, structures, and behaviors offer a playbook for balancing the sustainability benefits and costs of AI to capture its full potential.

1. Shapers move beyond compliance, using AI to unlock lasting sustainability impact   

With more than half of companies still in the exploring or piloting stages, there’s much to learn from those further along. Shapers take a fundamentally bolder approach to selecting, adopting, and getting value from their AI use cases. They prioritize applications with greater potential for long-term impact and enterprise value, such as scenario and risk modeling and sustainable product design—even if the path to returns is slower (see Figure 3).

Here are four use cases shapers are adopting that focus on both business and environmental goals:

• Long-term scenario and risk modeling. AI helps companies manage the increasing complexity of climate volatility, regulation, and technological change by processing vast data sets for more resilient, forward-looking planning. A global logistics firm, for example, uses AI to model geopolitical and climate disruptions—guiding infrastructure investment, fleet design, and contingency planning.
• Sustainable product and service design. As demand grows for greener products, companies must innovate without compromising on cost or performance. AI can support this by enabling generative design, material optimization, and tailored solutions. By employing an R&D digital twin to test ingredients and packaging for its cleaning and personal care products, a leading consumer goods company has been able to not only cut waste but also accelerate product innovation.
• Energy efficiency. AI supports decarbonization by analyzing complex operational data to pinpoint emissions hot spots and optimize performance. Real-time energy monitoring, predictive maintenance, and load balancing support faster, smarter decisions and measurable efficiency gains. In Singapore, the DecarboniSME gen AI platform is guiding small and midsize businesses through emissions tracking and tailored action plans—including solar installation and energy audits that have cut energy use by up to 10%.
• Green market opportunity identification. AI identifies high-potential markets and buyers for green products by analyzing customer behavior, sustainability goals, and competitive signals. First Abu Dhabi Bank leverages public data and large language models to prioritize sustainable finance opportunities, advancing its commitment to lend, invest, and facilitate $136 billion toward environmental and socially responsible solutions by 2030.

2. Shapers see the risks that others miss—and manage them better

Executives broadly agree that AI introduces new risks to sustainability, but approaches to address those risks vary widely. Shapers are four times more likely than laggards to perceive high sustainability-related risks from AI. Unlike laggards, they don’t just focus on compliance, privacy, and data security. They look further ahead to the disruptive and still-emerging risks AI poses to their teams (see Figure 4). For example, as AI reshapes roles and requires new skills, employee trust and engagement become critical. If these are not addressed, massive disruptions in productivity, morale, and organizational performance are likely.

Shapers are twice as likely to consider sustainability risks when making AI decisions and twice as likely to involve sustainability leaders in AI governance. They hardwire sustainability into AI decision making and governance by embedding it into model design, vendor selection, and oversight processes.

Figure 4

A leading engineering and infrastructure firm is using these types of approaches. Its AI-powered performance management tool that delivers personalized career growth and training recommendations has boosted employee engagement by a factor of five. At the same time, the company is using AI to improve its overall HR function, with AI-powered HR assistants and HR leadership assistants dramatically accelerating analyses and reducing both ticket volume and resolution time. 

3. Shapers build an integrated operating system

Most companies are starting to install the basic hardware for AI in sustainability, including data infrastructure, governance, and reporting. Shapers are building something more sophisticated: an integrated operating system.

This “OS” embeds sustainability into the way AI is selected, deployed, and governed. It connects internal users and external partners to enable smarter decisions, faster adoption, and broader innovation (see Figure 5).

Figure 5

Shapers focus on three actions to scale sustainable AI effectively:

• Program the core logic. Shapers hardwire sustainability into their AI strategy—targeting high-impact use cases tied to material ESG topics, even when return on investment takes time. They also reduce AI’s own footprint by setting energy limits, choosing greener models, and optimizing infrastructure.
• Configure for internal users. As with any good OS, the approach to AI in sustainability must work for users. Shapers upskill employees on sustainable AI and involve them in determining how it affects their roles. This builds trust, accelerates adoption, and ensures the system can flex, scale, and perform.
• Expand the network. The most effective systems are open. Shapers build external coalitions, partnering with start-ups, researchers, and policymakers to reduce risk, accelerate innovation, and influence the ecosystem around them.

The path forward

AI has the potential to become one of the most powerful tools in the sustainability playbook. But turning that potential into progress requires more than pilots. It demands focus, scale, and systems built for impact. The leaders we call shapers show what’s possible. Now is the moment to learn from them, raise your ambition, and embed AI for sustainability where it matters most.