‍

Introduction‍

Two hundred and fifty years ago, the Industrial Revolution catalyzed a new era of productivity. Despite numerous quality of life improvements, this movement also led to greater use of natural resources including energy, land, water, and air, contributing to the severe climate and biodiversity crises we are battling today.

Today, we are experiencing another economic transformation: the so-called “AI Revolution.” Like the industrial processes before it, artificial intelligence (AI) has the potential to greatly enhance the human experience while also presenting significant challenges for our planet and society.

Trillium’s purpose is to activate our clients’ capital to advance humankind toward a global sustainable economy, a just society, and a better world. We believe that a community’s sustainability depends upon a balance between equity, ecology, and economy and that our efforts should contribute toward achieving this balance. No single action is likely to accomplish all of these goals and likewise, no single company or investable asset will perfectly embody our endeavors However, these values are the guideposts we use to evaluate whether our business practices, investment decisions, and advocacy align with our ambitions.

This philosophy is integral to how we assess the opportunities and risks associated with AI and other emerging technologies. Trillium analysts work together seeking to identify sustainable companies that minimize environmental impact while generating value through innovation. We also assess the AI ecosystem holistically, such as by investing in companies adding renewable capacity to the grid, or selling products that make data centers more efficient.

‍

What is AI, and what are its impacts?‍

Artificial Intelligence 101

AI technology has existed since the 1950s, evolving to become a relevant tool in nearly every sector. Traditional AI operates using predefined algorithms and rules to perform specific tasks based on these guidelines. Humans have trained AI to follow specific rules without creating new content, such as with spam filters, or voice assistants like Siri and Alexa. Generative AI (Gen AI) represents the next step in this evolution. This AI technique, which exploded into popular culture with the release of ChatGPT, is capable of producing new and original content based on learned patterns from existing data. Gen AI models rely on machine learning, which includes several subfields, such as neural networks, deep learning, and large language models. These areas aim to mimic how the human brain works and need substantial computing power to function. Data centers supply this necessary computing power.

With any new advancement in technology, unforeseeable and foreseeable risks and opportunities are bound to arise. Gen AI has many exciting positive applications, including in fields that support the United Nation’s Sustainable Development Goals like education, healthcare, and crisis response. We also acknowledge that Gen AI risks remain uncertain, and in collaboration with our investment and advocacy colleagues, we are engaged with management teams and industry experts to gain a better understanding of AI use and field any questions.

‍

Electricity Impacts

To meet AI’s appetite for computing power, data centers consume a growing and significant amount of electricity, straining electrical grids. Data centers’ electricity consumption is mainly driven by running thousands of servers responsible for computing and cooling electronics. The International Energy Agency (IEA) estimates that 2% of total global electricity demand in 2022 came from data centers, cryptocurrencies, and AI. The IEA estimates that the global electricity consumption of these technologies will increase to about 6% of total demand by 2026 — an increase equivalent to adding a nation the size of Sweden to the grid.

This statistic excludes AI’s energy-intensive supply chain. The electricity required by AI has already affected the sustainable metrics of Trillium’s portfolio companies. For example, Alphabet Inc is no longer maintaining operational carbon neutrality, while Microsoft’s emissions have grown 29% since 2020 due to AI-supportive data center construction. Trillium’s Advocacy team recently questioned Apple on how the company plans to meet its greenhouse gas emissions reduction goals now that OpenAI is part of its supply chain. At Trillium, we continue to closely monitor AI-related emission increases in our holdings.

Electricity demand that is met through fossil fuel-based energy will be particularly detrimental to the Earth’s climate and communities in which fossil fuels are combusted. The environmental impact of data centers is heavily influenced by the sustainability of the local grid. In major AI markets, generation is still reliant on fossil fuels. In the US, the world’s largest data center market, 60% of utility scale electricity was generated by fossil fuels in 2023, largely natural gas. In China, nearly 60% of electricity in 2023 was generated from coal. A risk is that the increased demand for electricity may perpetuate the use of inefficient fossil fuel assets. For more information on how fossil fuel combustion is contributing to climate change, read Trillium’s approach to investing in energy and power.

‍

Water Impacts

In addition to climate change impacts from fuel combustion, many power plants use large cooling towers that convert water from reservoirs into steam emitted into the atmosphere. Data centers also use freshwater to keep servers cool, because electricity moving through semiconductors continuously generates heat. This is often accomplished with cooling towers as well, contributing to steam-based water loss.

The amount of water used by AI is significant: a two-week training for Microsoft’s GPT-3 AI program, for example, consumed about 500,000 liters of freshwater, equivalent to manufacturing 370 BMW cars. The water consumption would have tripled if Microsoft conducted this training in one of its Asia data centers, which are less efficient than its US data centers. Alphabet Inc.’s US data centers alone consumed an estimated 12.7 billion liters or about 5,080 Olympic-sized swimming pools of fresh water in 2021 for cooling.

Water use is especially problematic when data centers are built in areas that are experiencing water stress, such as in Arizona. Despite the prevalence of water scarcity, data center operators are drawn to such places due to low energy costs. In 2021, about one-fifth of data centers were found to draw water from moderately to highly stressed watersheds, mostly in the Western US.

Water on earth is finite — we cannot increase water use infinitely without consequence. Prolonged water stress can have acute effects on public health, agriculture and food supply, economic development, and biodiversity. Oftentimes, marginalized communities are the most impacted when resources like water are under stress.

‍

We don’t know what we don’t know

While the environmental and social impacts of fossil fuel and water stress are well evidenced, there are unforeseeable risks associated with any new technological development.

With AI, we believe many of these unknowns relate to social risks around justice, data integrity, and safety, in addition to unintended or problematic uses of AI technology. These risks are more difficult to assess than questions around environmental impacts or revenue generation.

Trillium’s Investment and Advocacy colleagues are in ongoing dialogue with corporate management teams to gain confidence in their ability to successfully manage such emerging issues. Recently, we have been concerned that Alphabet’s board may not be providing sufficient oversight of emerging AI-related social issues. Our Advocacy team has engaged the company over its unsuccessful (in our view) implementation of its seven AI principles and the company’s ability to comply with child protection laws.

At health insurer Elevance, our advocacy, in conjunction with others, led the company to conduct a third-party health equity assessment using a civil rights lens, focusing on its artificial intelligence practices and other areas of its business. Building upon a long history of pressing tech companies to address their social and environmental impacts, we will continue to engage with our portfolio companies around AI-related issues as they continue to evolve.

‍

Investing in Solutions

Trillium recognizes that Gen AI is going to be part of our society going forward, despite environmental or social impacts. We believe that there is a responsible and more sustainable way to implement this technology and seek opportunities to enable this version of the future. This approach contrasts our view on the fossil fuel or tobacco industries, for example, which we do not invest in due to the lack of a sustainable path forward.

In Trillium’s strategies, we seek to identify technology companies that are more energy and water-efficient than their peers, as well as ancillary companies that help enable this efficiency. To complete our holistic view of AI issues, we also look for utility companies that can serve AI-fueled energy demand with new renewable capacity, at scale. Regardless of sector, we are assessing if and how companies incorporate AI into their businesses, and whether that AI use is managed responsibly.

‍

Technology Solutions

At Trillium, we seek to invest in responsible companies that create infrastructure to enable Gen AI technology, rather than investing in more speculative new Gen AI companies themselves.

For example, leading technology company Nvidia has pioneered the Graphics Processor Unit (GPU), which has become the backbone of Gen AI progress. We also seek out companies that we believe make AI infrastructure more sustainable, through both hardware and software. An example in AI-related hardware is Allegro Microsystems, a company that develops less energy- intensive semiconductor technology and power integrated circuits. On the software side, Cadence offers products designed to make Central Processing Units (CPUs) and data centers more efficient. Confluent is another company that solves how to move large amounts of data, using less computing power. We believe these companies help address how we can minimize resource use as we adopt new technologies.

Trillium has longstanding relationships with companies like Alphabet and Microsoft that are expanding the use of Gen AI in consumer-facing products. For these companies, Trillium is keeping a close eye on responsible AI governance and how AI is affecting climate transition plans. We have been able to leverage our Advocacy relationships at these companies to engage on the responsible use of AI and will continue to do so as issues evolve.

‍

Industrial Solutions

In addition to hardware and software, companies can use energy-efficient industrial products and services when they build future data centers. These products help enable more sustainable cooling, control, and power management systems. Several of Trillium's industrial portfolio companies help provide data centers with more efficient technology. Within the heating and cooling space, Trane and nVent provide several sustainable cooling and data center efficiency solutions. Siemens AG and Eaton Corporation provide data centers with building and power management software solutions that allow them to run more efficiently. Across the industrials sector, Trillium analysts assess how companies are capitalizing on sustainable opportunities.

‍

Utility Solutions

Despite more energy & water efficient solutions, AI is still driving an increase in energy demand. Even so, AI is driving an increase in energy demand. To meet this demand, utilities risk slowing the pace of their low-carbon transition plans, keeping old, fossil fuel infrastructure on the grid longer than planned. Recently, a Kansas City area utility announced plans to delay retiring a coal plant that’s been operational since the 1960s in order to provide enough electricity to power
a new data center. This utility, along with other power companies, has also requested permission to build new gas-powered plants, locking in future emissions for decades. Trillium opposes such plans. Under our energy policy we will not invest in utility companies that are planning to invest in new fossil fuel-based power generation.

Instead, we look to invest in companies that have a proven track record of bringing new renewable capacity to the grid. One such company is NextEra, that we believe is a rare utility with a credible plan to succeed in a low-carbon economy and is helping facilitate a just energy transition across the US. The company continues to meet Trillium’s fundamental sustainability criteria in addition to Trillium’s Energy and Power specific policy, which we initially published in 2021.

NextEra is a world leader in renewables and battery storage with 34 GWs of renewables in operation, 20 GW+ in backlog, a 300 GW renewable pipeline, and 20 years of development expertise. FPL, NextEra’s Florida-regulated utility has already transitioned away from coal & oil, shifting toward renewables whilst keeping customer bills 37% below the national average. We acknowledge NextEra owns some legacy natural gas and nuclear power plants but has no plans to add new fossil fuel or nuclear plants.

Ormat is another portfolio name that is fueling the low-carbon energy transition, by bringing new geothermal energy to the grid. Ormat’s geothermal plants operate without fossil fuel consumption and use less water than conventional geothermal steam turbines. Geothermal energy has a unique role to play, as it can replace baseload capacity that is typically generated through fossil sources. Both NextEra and Ormat are working with technology companies and data centers on clean energy solutions.

‍

Conclusion

While many of the risks and opportunities around Gen AI are apparent, we know others will emerge as the technology matures. Trillium’s teams will remain flexible as we continue to learn more about this technology and its industrial capabilities. Until then, we remain committed to assessing Gen AI companies, and all companies, through a lens of economic advancement, environmental sustainability, and social justice.

‍

‍

‍

IMPORTANT DISCLOSURE

This information is not intended to be a forecast of future events or a guarantee of future results. These views may not be relied upon as investment advice. The information provided in this material should not be considered a recommendation to buy or sell any of the securities mentioned. It should not be assumed that investments in such securities have been or will be profitable. The specific securities were selected on an objective basis and do not represent all the securities purchased, sold or recommended for advisory clients. The information contained herein has been prepared from sources believed reliable but is not guaranteed by us as to its timeliness or accuracy and is not a complete summary or statement of all available data. This piece is for informational purposes and should not be construed as a research report.

‍

‍

‍

‍

‍

‍