Using natural gas to power datacenters:
Environmentally reckless & systemically negligent
By David Kyler, Center for a Sustainable Coast
As Georgia continues using natural gas to produce electricity at new and existing powerplants, the legitimacy of this policy deserves greater scrutiny. Burning natural gas to generate electricity – now mainly needed for AI and datacenters – is based on dubious claims asserting lower carbon emissions than burning coal and greater round-the-clock reliability than using solar or wind power. Recent research and tech innovations challenge both these claims.
According to Environmental Defense Fund, U.S. natural-gas pipelines leak more methane than previously estimated. Those leaks have the same near-term heat-trapping impacts as driving approximately 50 million passenger vehicles. Plus, there are substantial methane leaks at homes, businesses, and factories. Combined, all these leaks, above and below ground, are disturbing.
Because methane (CH4) has far worse environmental effects than carbon dioxide (CO2), even nominal natural-gas leakage significantly worsens climate-change damage. Per ton of emissions, methane has about 80-times the heat-trapping effects as comparable amounts of CO2. Energy experts advise that leaks greater than 2% make using natural gas more harmful than burning coal.
Contradicting this often-ignored problem are breakthroughs in battery technology that enable intermittent energy sources of power from sun and wind to support energy needs day-and-night, throughout all weather conditions. The latest iron-air batteries being used at a Google datacenter in Minnesota have a recharge life of 100 hours, with an output of 300 megawatts of power. Supported by this battery backup, the entire facility will be operated using only solar power, at lower cost than burning natural gas.
This flagrant indifference toward environmental damage caused by conventional energy and development policy is worsened by recent news that the Trump administration is eliminating policies that reward energy efficiency, the transition from fossil-fuels to EVs and clean energy, and incentives for buying electric appliances that can be powered by clean energy.
Moreover, according to recent analysis, using natural gas is also limited by existing power-grid capacity and constraints of hardware performance:
“… around 30% of the power flowing into data centers is not used to generate AI, according to [AI industry leader] Nvidia. Much of it gets used up by cooling systems that keep servers from overheating and by electricity traveling long distances…. That is amplifying carbon emissions from datacenter energy use given operators are increasingly relying on natural gas and coal-fired plants to power their projects. Microsoft Corp., for example, is considering whether to delay or abandon its ambitious clean-energy targets as it tries to remove hurdles that could hold it back in the AI race….”
[https://www.bloomberg.com/graphics/2026-ai-data-center-redesign/]
Nvidia is seeking to save energy by improving the efficiency of delivering electricity from the grid to datacenter microchips. Electricity enters data center facilities from the grid using alternating current (AC), the most efficient way to transport massive amounts of power over long distances. But in order for chips to use it, the power must be converted to direct current electricity (DC), which provides a consistent flow of power, enabling proper operation.
Datacenters also drastically reduce the voltage of grid power from 34,500 volts AC, dangerously high levels of voltage carried by major powerlines, to the 12 volts DC required by microchips. This power conversion takes numerous steps, each losing efficiency through emitted heat, consuming energy for cooling needed to protect microprocessor performance. Cumulatively these inefficiencies amount to substantial energy demands with costly financial and environmental consequences.
The self-defeating nature of using natural gas or any other dirty-energy fuel to power advanced technology is both ironic and perilous. It is a prime example of the tragic failure to honor the groundrule that systemic problems require systemic solutions. Datacenters, AI, disinformation, energy and development policy, food supplies, public health, environmental quality – including climate change, life-system viability, and toxic contamination – are interlocked issues, each profoundly affected by the others in ways that are blatantly ignored in public policy.
Federal and state laws and budgeting decisions reflect alarmingly fragmented thinking that cannot possibly serve the long-term public interest. Attempting to resolve these interconnected perplexing problems individually is inherently self-defeating and diversionary.
Reducing the destructive impacts of datacenters requires using clean energy, meaning that energy policy must no longer subsidize fossil fuels. Likewise, disinformation spread by social networking and AI algorithms must be curtailed by strictly regulating high-tech applications, which – in turn – depend upon well-regulated datacenters.
To serve our common interests, citizens must advance a unified campaign addressing an array of interlinked social, scientific, political, environmental, and technological factors. Fragmented approaches to these complex problems must be replaced by well-integrated systemic policies.
Georgians deserve public policy based on systemic analysis and scientific facts, not outdated rationalizations.