Artificial Intelligence Versus Reality: The ESG Risks Behind the AI Boom

Key Insights:

Investors continue to demonstrate interest in the links between sustainability and artificial intelligence (AI). As data centers scale, investors are demanding increased clarity about the credibility of Big Tech climate strategies, especially amidst the uncertain implementation of the EU AI Act and the potential social ramifications of AI adoption.

This article explores the concerns that investors have about the credibility of companies’ climate strategies, as well as the environmental, social and governance risks associated with rising AI demands.

Investors Question Credibility of Big Tech’s Climate Strategies

Investors are entering into a more mature stage of climate scrutiny. At first, investors demanded that companies set climate targets. Now that most companies — especially tech companies — have them, the concern has shifted. Now, investors want to know whether those targets still make sense as business models evolve. This is especially true within the context of AI, with data centers driving a significant increase in electricity demand.¹

In the cases of Amazon, Meta, and Alphabet — three companies at which climate-related shareholder proposals were filed — investors asked for clarity on the execution of these companies’ climate strategies,² rather than for more ambition on paper. These proposals also reflect a growing awareness of the limits of current accounting practices. Annual matching³ and market-based reporting⁴ can create the impression of smooth climate progress, even when data centers are located in regions that still rely heavily on fossil fuels.

Ultimately, climate credibility is transforming into a governance issue. With energy use growing faster than clean energy can be added to the grid,⁵ investors are requiring companies to explain how their plans meet existing commitments, whether they’re credible in real-world conditions, and how they translate into real emissions reductions. Given the scale of the challenge ahead, strategies need to be resilient, transparent, and realistic.

What Does Climate Credibility Mean to Investors as AI Demand Grows?

In the past, tech companies were able to reasonably commit to net zero pathways based on assumptions of steady business growth and ongoing efficiency improvements. However, given the sheer scale and pace of AI-driven expansion, this is no longer the case. Data center electricity demand is rising quickly, and risks outstripping efficiency gains and renewable energy procurement strategies. This creates a growing gap between long-term climate commitments and short-term operational realities. 

From an investor perspective, credibility means three things:

1. Planning realism: are companies stress-testing their climate pathways against high-growth scenarios for AI and cloud computing? 
2. Operational integrity: are companies relying mainly on accounting tools like renewable energy certificates, or are they actually driving new clean generation where and when their data centers operate?
3. Transparency: are companies clearly disclosing their renewable energy sourcing practices, and are they making both market-based and location-based emissions data available in their mainstream reports, so investors can have an understanding of what’s really happening on the ground? 

In the tech sector, credible climate strategies will start to depend less on headline commitments and more on how companies manage this energy transition in practice. This includes how they shape electricity markets, invest in grid decarbonization, and adapt their strategies as energy constraints become more binding.  

How Will Companies Adapt to Increasing Energy Demand Fueled by the AI Boom?

Current trends — such as an increase in renewable energy and alternative energy sources like nuclear power — will likely continue in 2026. Specifically, there has been a resurgence in the use of nuclear power, and Meta, Alphabet, and Amazon all signed onto nuclear deals to help run their data centers in recent years.⁶ Big Tech companies are shifting towards sourcing and building their own dedicated power supplies, emphasizing how immense the power demand is for these companies and their data centers.

Sustainalytics expects to see continued reliance on market-based instruments such as renewable energy certificates (RECs)⁷ and virtual power purchase agreements (VPPAs).⁸ Increased investment from Big Tech into carbon capture⁹ and carbon removal technology can also be noted.¹⁰ Specifically, Microsoft has made ambitious commitments on the carbon removal front.¹¹ If it can make significant progress and succeed here, other firms may follow suit.

However, rising energy demand and generation due to the powering of data centers is leading to a significant increase in carbon emissions. As a result, companies are struggling to achieve greenhouse gas (GHG) reduction and meet net zero targets. In some cases, targets have been rolled back entirely.¹²

While renewables, nuclear energy, and RECs can help supply this rising energy demand, reaching global climate targets requires companies to also vastly reduce their absolute emissions. At this point, this is not something that we’re seeing from Big Tech.

What Water-Related Risks are Data Center Operators Exposed to?

As data center cooling continues to require a greater volume of water, and as water consumption increases globally, water shortages are becoming a material risk to Big Tech and data center providers. Not having enough water to efficiently cool data centers could lead to operational impacts.  

Additionally, many data centers are constructed in water-stressed regions and areas prone to droughts. For example, in the US, Texas, California, and Arizona all house a large number of data centers. While these hot and dry climates are attractive for the solar energy that is often used to help power these data centers, buildouts in water-stressed locations are putting extra stress on local water basins.¹³

This increased demand for water also leads to a social risk: community opposition of data center buildouts. Residents are worried about access to water and potential price increases when data centers are constructed in their communities. Essentially, locals and data center operators are fighting for the same water resources. In the US, local opposition led to the blocking or delay of USD 98 billion in data center projects in just Q2 2025.¹⁴ In the year prior, USD 64 billion worth of data center projects were blocked or delayed.¹⁵

Regulatory Flux and AI Governance

Uncertainty around the implementation of the EU AI Act could play a significant role in governance risk related to AI in 2026. The world’s first comprehensive AI regulatory framework, the AI Act, categorizes different levels of risk in AI systems: unacceptable, high, limited, and minimal. 

Systems classified as high risk are those that can significantly affect people’s health, safety, or fundamental rights. The act places strict requirements on both providers and users of high-risk AI systems, which could impact companies operating in sectors like finance, human resources, healthcare, education, and critical infrastructure. 

Some of the obligations of the EU AI Act include:

• Robust risk management systems 
• Quality and governance of training data
• Extensive technical documentation
• Human oversight
• Traceability
• Explainability of decision making
• Incident reporting and conformity assessment procedures

Providers must also register these systems in a central EU database and maintain cybersecurity and accuracy throughout the entire lifecycle of the system. 

The act entered into force in 2024 and is now being phased in, with first requirements becoming applicable in August 2026, and additional ones coming into force through 2027.¹⁶ Non-compliance could lead to fines, sanctions, or restrictions on the use of AI systems in the EU market.¹⁷

Currently, the uneven and delayed rollout of the full set of requirements for high-risk AI has created uncertainty for companies and investors around how they will be implemented, enforced, and audited across different jurisdictions and sectors. Lobbying from several actors, including large technology firms, AI company founders, the US government, European industry groups, and others looking to ease compliance, has also added to this uncertainty.¹⁸

For Big Tech companies, AI governance is critical. Companies that embed good governance early on are typically better prepared for tightening regulations. On the other hand, laggards tend to face higher compliance risks, reputational damage, litigation exposure and, in some cases, even exclusion from public procurement or sustainable investment portfolios.¹⁹

The Social Impact of AI Adoption

In 2026, it is likely that the social dimension of AI will become more material for investors, especially in terms of workforce transitions. Over recent years, there has been a jump in how often companies talk about AI, with disclosures rising from around 4% in 2020 to 43% by 2024.²⁰ Investors increasingly want to understand what AI adoption means for the workforce and how companies are managing this transition. 

Certain sectors such as information, financial activities, and professional services could potentially be heavily impacted. Since 2022, these sectors have seen some of the most significant changes in their occupational mix,²¹ and they’re also more vulnerable to automation because they have high exposure to AI, and a large share of clerical, administrative, and routine cognitive roles. 

In the short-term, we expect to see automation boost productivity, although it could also lead to workforce reductions, temporary increases in unemployment, and greater income polarization. Women are expected to be more affected, mostly because of their high representation in clerical and administrative roles.²² Without intervention, AI could reinforce existing inequalities. 

Companies could mitigate these risks by adopting hybrid human–AI models. This means using AI for automation and augmentation, while shifting human work toward more analytical, creative, relational, and supervisory roles. When done well, hybrid models can improve job quality and help preserve institutional knowledge, instead of just replacing workers. Additionally, leading companies are investing in workforce transitions by closing skills gaps, redeploying employees, and maintaining morale during periods of transition. 

From an ESG perspective, these social impacts translate directly into reputational, regulatory, operational, and financial risks. Poor management of these risks can lead to fines, litigation, loss of stakeholder trust, and reduced investor confidence. 

To learn more about how sustainable investing trends are evolving, read Sustainalytics’ full report, Sustainable Investing Trends to Watch in 2026, or watch our on-demand webinar.

---

References

1. International Energy Agency. “Energy and AI.” 2025, https://iea.blob.core.windows.net/assets/601eaec9-ba91-4623-819b-4ded331ec9e8/EnergyandAI.pdf.
2. Felleca, M. 2025. “Can Big Tech Keep Its Climate Commitments as Data Centers Scale?” Morningstar Sustainalytics. December 8, 2025. https://www.sustainalytics.com/esg-research/resource/investors-esg-blog/can-big-tech-keep-its-climate-commitments-as-data-centers-scale.
3. Annual matching is a method in greenhouse gas and energy-attribute accounting, whereby an organization's energy consumption over a full year is matched with an equivalent amount of renewable energy generation — also measured on an annual basis — using renewable energy certificates or power purchase contracts.
4. Market-based reporting is a method of scope 2 accounting whereby an organization calculates its electricity-related emissions using emission factors from the specific energy products it purchases, such as power purchase agreements and renewable energy certificates. Emissions reflect the attributes of the electricity the organization chooses to buy.
5. Poole, K., Paris, A., & Fisher, J. 2025. “Compute and consequence: AI energy demand in a rapidly evolving grid landscape.” As You Sow. https://static1.squarespace.com/static/59a706d4f5e2319b70240ef9/t/68c08723c6524e7b42a8e9ef/1757447971180/AsYouSow2025_Compute-Consequence_Final.pdf.
6. Clancy, H. 2025. “7 companies helping Amazon, Google, Meta and Microsoft go nuclear.” Trellis. June 12, 2025, https://trellis.net/article/amazon-google-meta-and-microsoft-go-nuclear/.
7. Renewable Energy Certificates represent proof that 1 megawatt-hour (MWh) of electricity has been generated from renewable sources and added to the power grid.
8. Virtual Power Purchase Agreements are contracts between a buyer and a renewable energy producer, in which the buyer commits to buying a variable amount of renewable energy over a set period.
9. Carbon capture involves capturing carbon dioxide, either at the source of emission, or directly or indirectly from the atmosphere. This carbon dioxide is then typically transported and stored or buried deep underground.
10. Green, A. 2025. “Big Tech Firms Microsoft (MSFT) and Alphabet (GOOGL) Lead in Durable Carbon Removal Investments Exceeding $10 Billion.” Carbon Credits. November 21, 2025. https://carboncredits.com/big-tech-firms-microsoft-msft-and-alphabet-googl-lead-in-durable-carbon-removal-investments-exceeding-10-billion/.
11. Microsoft. “Carbon dioxide removal.” Accessed February 2, 2026, https://www.microsoft.com/en-us/corporate-responsibility/sustainability/carbon-removal-program.
12. Pucker, K. 2024. “Companies Are Scaling Back Sustainability Pledges. Here’s What They Should Do Instead.” Harvard Business Review. August 20, 2024/ https://hbr.org/2024/08/companies-are-scaling-back-sustainability-pledges-heres-what-they-should-do-instead.
13. United Nations University. 2026. “World Enters ‘Era of Global Water Bankruptcy’ UN Scientists Formally Define New Post-Crisis Reality for Billions.” January 20, 2026. https://unu.edu/inweh/news/world-enters-era-of-global-water-bankruptcy#:~:text=%E2%80%9CThis%20report%20tells%20an%20uncomfortable,shocks%20that%20can%20be%20overcome.
14. Data Center Watch. “Q2 2025 UPDATE: 125% Surge in Data Center Opposition.” 2025. https://www.datacenterwatch.org/q22025#:~:text=Key%20Takeaways:,to%20data%20centers%20is%20accelerating.
15. Data Center Watch. “$64 billion of data center projects have been blocked or delayed amid local opposition.” 2025. https://www.datacenterwatch.org/report.
16. European Parliament. 2025. “AI Act Implementation Timeline.” October 6, 2025. https://www.europarl.europa.eu/RegData/etudes/ATAG/2025/772906/EPRS_ATA%282025%29772906_EN.pdf.
17. European Parliament. 2025. “EU AI Act: first regulation on artificial intelligence.” February 19, 2025. https://www.europarl.europa.eu/topics/en/article/20230601STO93804/eu-ai-act-first-regulation-on-artificial-intelligence.
18. Moens, B. 2025. “EU set to water down landmark AI act after Big Tech pressure.” Financial Times. November 7, 2025. https://www.ft.com/content/af6c6dbe-ce63-47cc-8923-8bce4007f6e1.
19. Arga e Lima, F. 2025. “Navigating the EU AI Act: A Comprehensive Analysis and Compliance Guide.” LEXR. November 18, 2025. https://www.lexr.com/en-de/blog/navigating-the-eu-ai-act-a-comprehensive-analysis-and-compliance-guide.
20. Lucas G. et al. “Are Companies Taking AI Risks Seriously? A Systematic Analysis of Companies’ AI Risk Disclosures in SEC 10-K Forms.” Maastricht University Law and Tech Lab. 2025. https://arxiv.org/abs/2508.19313.
21. Gimbel, M, et al. 2025. “Evaluating the Impact of AI on the Labor Market: Current State of Affairs.” The Budget Lab. October 1, 2025. https://budgetlab.yale.edu/research/evaluating-impact-ai-labor-market-current-state-affairs.
22. Goldman Sachs. 2025. “How Will AI Affect the Global Workforce?” August 13, 2025. https://www.goldmansachs.com/insights/articles/how-will-ai-affect-the-global-workforce.