AI Race Forces Google and Tech Giants to Invest in Natural Gas Plants in the US

The unbridled pursuit of leadership in artificial intelligence is forcing tech giants to rethink their energy supply strategies. A new data center campus, Goodnight, located in Armstrong County, Texas, and funded with investments from Google, illustrates this growing conflict. The project plans to use private natural gas turbines to power part of its operations, which, according to estimates based on environmental permit applications, would result in the emission of 4.5 million tons of greenhouse gases annually. This volume of pollution is equivalent to the environmental impact of adding nearly 970,000 combustion-engine cars to the streets, even exceeding the emissions of a conventional coal-fired power plant.

The impasse between climate goals and AI demand

For years, companies like Google were celebrated by environmental groups for their public commitment to renewable energy. However, the current landscape reveals a deep tension: the infrastructure required to support large-scale AI model processing requires an energy load that current power grids cannot supply with the desired speed. Michael Thomas, founder of Cleanview, points out that the Texas project is a clear indicator that, under the pressure of the AI race, the sector is prioritizing immediate energy availability over decarbonization promises, resorting to fossil fuels through power plants installed directly at the site of demand.

Technical specifications and reliance on gas

The Goodnight campus, which is being built by the AI infrastructure firm Crusoe, features a hybrid energy configuration. Planning indicates that, of the six planned units, the first four will be connected to the public power grid. However, the two remaining units will be powered by a dedicated gas plant, operating under the so-called behind-the-meter model. The project also includes 265 megawatts of wind power, according to interconnection agreements filed with the Public Utility Commission of Texas. Although Google has stated that it does not have specific contracts for the use of gas at this facility, the presence of environmental permit applications detailing massive emissions suggests that the infrastructure is being prepared for this fossil fuel dependency.

Systemic impact on the tech sector

This move is not an isolated case, but rather a trend spreading across the United States. The Global Energy Monitor reported that there are about 100 gigawatts of natural gas-powered energy capacity in development in the country for the sole purpose of powering data centers. The strategy of building private generators is a direct response to long wait times for connection to the national power grid and the fear that massive energy consumption by data centers will increase electricity bills for ordinary consumers. Cully Cavness, co-founder of Crusoe, defends this approach as a pragmatic step, describing gas as a "necessary bridge" while solutions such as batteries, wind, and small modular nuclear reactors do not yet reach the necessary scale.

Competitive landscape and rising emissions

The Goodnight project, despite its scale, is not the most polluting one under development. The sector is watching other ventures with caution, such as "Project Jupiter," a partnership between OpenAI and Oracle in New Mexico, which estimates emissions of up to 14 million tons of greenhouse gases per year. Meanwhile, Microsoft has also formalized partnerships with oil giants, such as Chevron, to secure up to 2.5 gigawatts of gas power for its operations in Texas. These moves show that the AI "arms race" is forcing a strategic alliance between the tech sector and the fossil fuel industry, challenging the narrative that digital progress is inherently sustainable.

The future of energy infrastructure

In the medium term, the model of self-generation of energy is expected to become the de facto standard for the development of large-scale data centers. The speed of implementing these plants, compared to the years of bureaucracy required to expand regional power grids, makes natural gas the most attractive option for companies. Google, while continuing to invest in solar and wind energy projects, finds itself in a delicate position, balancing the need to reduce its carbon footprint—which, according to recent reports, has increased by nearly 50% over the last five years—with the insatiable technical demand of its AI services. The future points to a scenario where the energy efficiency of algorithms will have to evolve as fast as energy generation capacity, at the risk of corporate sustainability goals becoming obsolete in the face of the infrastructure reality required to sustain the advancement of Artificial Intelligence.