By Olivia Raimonde
Anthropic PBC says it expects to face heightened scrutiny of how it tackles environmental and social tensions associated with the build-out of AI infrastructure.
The company, which filed for an initial public offering on Monday, said it was aware that its efforts to secure power may face regulatory hurdles if there is a perception that its activities are affecting utility costs or the environment more broadly.
“Data center development has prompted public controversy and regulatory scrutiny regarding environmental impact, land and water use,” the company noted in its IPO filing. At the same time, Anthropic’s “brand is subject to heightened scrutiny with respect to perceived societal, environmental, labor, health, and safety-related impacts,” it also said.
The comments come as companies building out the infrastructure needed to power artificial intelligence face increasing pushback from local communities worried about the impact on precious resources. Last month, Oracle Corp. cited force majeure after opposition from environmental nonprofits and regulatory setbacks complicated efforts to secure power for a data centre campus in New Mexico. That project is intended to provide computing power for OpenAI.
Also Read
In its S-1 IPO filing on Monday, Anthropic also highlighted climate change and environmental concerns more broadly as risk factors.
“The physical effects of climate change may increase our operating costs over time, as higher average temperatures can increase the power consumption and water usage required to cool equipment at the facilities on which we depend,” the company said.
AI’s risks and opportunities “are now core financial and fiduciary considerations, not side issues,” said Blaine Townsend, director of sustainable, responsible and impact investing at Bailard, a wealth and asset management firm. “What we’d want to see, industrywide, is disclosure standards that keep pace with that reality, not just risk factors written to protect against liability, but a real accounting of environmental and social externalities.”
Bailard’s research has highlighted three specific categories in which AI’s environmental footprint is material: data centres disproportionately located in communities already burdened by pollution, AI’s water use and its energy demand, Townsend said. “When we evaluate an AI company’s disclosures, we look for whether they address each of these three categories specifically.”
Artificial intelligence companies have come under scrutiny from activist groups, community members, regulators and investors. Some states have pushed for moratoriums on new data centre builds as they assess the potential impact on the environment and power prices. Anthropic acknowledged in its S-1 that this backlash could result in reputational harm to its business.
The Anthropic filing contrasts with SpaceX’s S-1, which had no mention of climate change. SpaceX subsequently amended its S-1 to add water scarcity as a risk factor.
Anthropic also referred to “increasing scrutiny” related to its ESG practices and reporting “both in the United States and internationally”. And it mentioned a “particular focus on concerns relating to AI and its impact on society and the environment”, including the power-intensive nature of its business, public concern about water usage and reliance on critical minerals.
“Some investors may use these non-financial performance factors to guide their investment strategies and, in some cases, may choose not to invest in us if they believe our policies and actions relating to ESG are inadequate,” it said.