Natural Gas Becomes Key Factor in Generative AI Expansion Plans

Anthropic's S-1 filing has shed light on the growing reliance of generative AI on natural gas, a fossil fuel primarily composed of methane, highlighting the sector's vulnerability to rising fuel costs and supply chain constraints.

The company acknowledges that its data center buildout is becoming increasingly dependent on natural gas, which is expected to increase operating costs due to rising global temperatures caused by burning fossil fuels. This has significant implications for Anthropic's expansion plans in Texas and New York, where it aims to establish new data centers.

Anthropic leases capacity from hyperscalers but also plans to build its own data centers, which will rely on electricity generated by natural gas turbines on-site. The company faces intense competition for access to electrical supply and data center locations with adequate power availability.

The US is planning to build more fossil fuel generation capacity than in a decade as data centers demand increasing amounts of electricity. This shift towards fossil fuels has been facilitated by the previous administration's regulatory policies, which have tilted the landscape in favor of natural gas over renewable energy sources like solar and wind.

The expansion plans are being met with growing concerns about the environmental impact of relying on fossil fuels to power data centers, particularly as global temperatures continue to rise due to climate change.

A shortage of gas turbines has begun to slow down data center construction projects worldwide, as developers struggle to secure the necessary equipment for their facilities.

In an effort to mitigate this issue, some companies are opting to install gas engines directly on-site, rather than waiting for extended periods to connect to existing power grids. However, this approach is resulting in less efficient and more polluting on-site power generation due to the use of simpler turbine technology.

A recent study conducted in Texas found that increasing electricity demand from data centers could lead to a significant deterioration in air quality over the next five years. The researchers estimated that this trend could undo decades of progress made in reducing pollution levels, resulting in higher concentrations of soot and smog-forming pollutants as well as carbon dioxide emissions.

As data center operators continue to rely on fossil fuels to power their operations, they are also facing rising costs due to increasing competition for natural gas and pipeline capacity. This is particularly concerning for companies like Anthropic that have highlighted climate change as a significant risk to their bottom line in their recent S-1 filing.

The reliance on gas power generation is further complicated by the fact that it currently accounts for 40 percent of the US electricity mix, with Americans already feeling the strain of rising energy costs due to utilities scrambling to upgrade infrastructure and meet growing demand.

The construction of data centers has been met with resistance from local residents across the US, resulting in dozens of proposed projects being blocked.

This opposition has significant implications for companies like Anthropic, which relies on large-scale computing infrastructure to power its AI models. To mitigate these challenges, Anthropic is exploring alternative energy sources, including gas-powered generation, to supplement its electricity needs.

The company acknowledges that its efforts to secure electrical power may be impacted by concerns over rising utility bills and the strain they place on the broader grid. If regulators impose conditions or charges to address these impacts, it could raise Anthropic's operating costs and potentially delay or abandon planned data center projects.

Ultimately, these challenges threaten to hinder Anthropic's growth, as any increase in energy or compliance costs could limit its ability to scale its compute infrastructure and serve its customers.