L&G Presses Meta for Clearer Climate Plan as AI Power Demand Grows

Legal & General’s asset management business is pressing Meta Platforms for more detail on how the technology company intends to reconcile the rapid expansion of artificial intelligence infrastructure with its climate goals, as data-center electricity demand becomes an increasingly important issue for investors.

The concern forms part of L&G’s engagement with companies across the AI data-center value chain. Since 2025, its investment stewardship team has been engaging with hyperscale technology companies, semiconductor manufacturers, utilities and real estate businesses to assess their exposure to the energy and environmental risks associated with the growth of artificial intelligence.

In its latest engagement disclosures, L&G said Meta could provide clearer information on how it balances the need for rapid access to electricity, often described as "speed to power", with the requirement for clean and reliable energy. The investor also wants greater clarity on the role natural gas could play in Meta’s electricity supply and on measures to reduce Scope 3 emissions from capital goods such as servers, equipment and data-center infrastructure.

The issue has become more significant as Meta accelerates investment in computing capacity. The company said in July that it expects capital expenditure of between $130 billion and $145 billion in 2026, including principal payments on finance leases, largely to support AI initiatives and its core business. Meta spent $31.08 billion on capital expenditure during the second quarter alone.

Investors Seek Greater Disclosure

L&G demonstrated its position at Meta’s annual shareholder meeting on May 27, when it supported a shareholder proposal requesting a report explaining how the company intends to meet its climate commitments as electricity demand from AI and planned data centers increases.

The proposal, filed by investors led by As You Sow and the Presbyterian Church (U.S.A.), argued that Meta’s growing infrastructure footprint could increase dependence on fossil-fuel generation and make its existing climate targets harder to achieve. It also cited concerns that rising electricity requirements could contribute to new gas-fired generation or delay the closure of higher-emitting power plants in some markets.

Meta’s board recommended voting against the proposal. The company argued in its proxy materials that its existing reporting, renewable-energy procurement and broader sustainability strategy already provided shareholders with sufficient information.

The proposal ultimately failed, receiving about 343 million votes in favor compared with approximately 4.61 billion against. Meta’s dual-class share structure gives its Class B shares 10 votes each, significantly influencing the company’s shareholder voting outcomes.

L&G nevertheless said it supported the proposal because the expansion of generative AI and large language models is increasing both electricity and water requirements. It warned that rapidly increasing demand from data centers could also influence utilities’ decisions about when to retire coal and gas assets, potentially locking in emissions for longer than previously anticipated.

Meta Expands Both Clean Energy and Gas-Backed Power

Meta continues to maintain substantial clean-energy commitments. The company says it has matched 100% of its annual electricity consumption with clean and renewable energy since 2020 and that projects it has supported are adding more than 30 gigawatts of clean and renewable generation to electricity grids globally. It also continues to target net-zero emissions across its value chain and water positivity.

The company is also becoming a major purchaser of nuclear power. Agreements involving Constellation Energy, Vistra, TerraPower and Oklo are intended to preserve existing nuclear generation and support new capacity. Meta said in January that its latest agreements could support as much as 6.6 GW of new and existing nuclear energy by 2035.

However, near-term electricity requirements are also increasing Meta’s exposure to natural gas. The company’s departure from the RE100 corporate renewable-energy initiative in July drew additional attention to the issue, particularly because Meta has supported gas-fired generation associated with major new AI data centers in the United States. Meta has continued to state that it will match its data-center electricity consumption with clean and renewable energy.

This distinction is becoming increasingly important for assessing the environmental footprint of AI. Annual renewable-energy matching can compensate for electricity consumption by supporting clean generation elsewhere or at different times, but it does not necessarily mean that a data center is physically supplied by carbon-free electricity during every hour it operates.

AI Infrastructure Becomes an ESG Issue

For investors, the debate goes beyond Meta. Artificial intelligence is creating a new category of climate and infrastructure risk for technology companies whose energy requirements are expanding faster than many electricity systems can add transmission and firm low-carbon generation.

Large data centers need substantial volumes of reliable electricity and can take less time to construct than new transmission lines, nuclear facilities or some renewable-energy projects. That mismatch is increasing pressure on utilities and technology companies to secure generation quickly, potentially making existing gas plants or new gas capacity attractive in the short term.

At the same time, investors are increasingly examining emissions beyond electricity consumption. Building data centers requires large quantities of steel, concrete, semiconductors, servers and electrical equipment, meaning Scope 3 emissions can rise even where operational electricity is matched with renewable generation.

Meta has acknowledged this challenge and says it is working on lower-carbon concrete, recycled metals and plastics, hardware circularity and other measures designed to reduce embodied emissions from its infrastructure.

The disagreement between L&G and Meta therefore illustrates a broader challenge facing the technology sector. Companies can continue expanding renewable and nuclear procurement while still creating additional fossil-fuel demand in specific locations as AI electricity consumption accelerates.

For investors, utilities and policymakers, increasingly important indicators will include not only annual renewable-energy purchases but also the location and timing of clean generation, additional fossil-fuel capacity required for new data centers, transmission investment and the embodied emissions associated with infrastructure construction.

As AI investment moves further into the physical energy system, those factors are likely to become increasingly important in assessing whether corporate net-zero strategies remain credible during a period of unprecedented growth in computing demand.

Source: www.bloomberg.com