2.1 Billion Metric Tons: U.S. Power CO2 Could Match Mexico

By Tom Whitfield

How large is the projected emissions increase?

A Sierra Club estimate, cited by Bloomberg, projects that pending U.S. data center buildouts will drive power-sector emissions to 2.1 billion metric tons of carbon dioxide by 2030, 42% above 2025 levels. The total could match Mexico's carbon dioxide output and undo 15 years of U.S. greenhouse-gas progress.

The figure covers the wider U.S. power sector once electricity demand from pending builds is included. It is not a facility-by-facility emissions inventory, and it does not assume that every proposed data center will reach operation.

The Sierra Club, one of the oldest environmental preservation organizations in the United States, frames the estimate as a warning. Principal advisor for climate and energy Jeremy Fusher said: “the numbers continue to sort of boggle our minds.”

The club's previous estimates ran from one-quarter to one-half of the current projection, a range that shows how much the forecast has moved since the prior work. The estimate remains tied to projects seeking approval and power, not emissions already produced. Turbine availability, grid access and project cancellations could all move the final total below this projection.

Can renewable-power rules keep pace?

More countries and states now require new data centers to obtain a large share of their power from green sources. Those laws rarely apply retroactively, and many include phase-in periods, so they shape new construction more directly than the installed base.

Ireland's rule shows the timing: new data centers must run on 80% renewable power only six years after switching on. Because the obligation starts after operation, a site can initially operate under a less restrictive power mix. The effect is a lag between opening day and the clean-power deadline.

Existing facilities also remain outside the rule when legislation applies only to new projects. In short, renewable mandates can push cleaner power later without changing near-term demand for reliable generation.

Why do gas turbines become the bottleneck?

Speed now governs data-center construction, and roughly 40% of U.S. data centers already rely on natural-gas turbines. The current buildout outlook assumes renewable additions will fall short of demand, pointing to more turbine reliance rather than an easing of the equipment shortage.

Data-center demand has absorbed turbine capacity much as it has consumed RAM, GPUs and storage. Turbine prices have risen 50%, production is backlogged for years, and one manufacturer says it will open production slots for 2032.

The 2032 slot falls beyond the estimate's 2030 endpoint, so the delivery schedule may not match the build schedule it is meant to support. That scarcity can shape which AI sites move from announcement to operation. Turbine availability now stands beside grid access as a practical limit on how many projects reach completion.

What could change the outcome?

Brendan Pierpont from Energy Innovation, another climate-focused organization, says the dangerous trajectory remains difficult to forecast. He cites turbine availability, grid-connection delays and other project delays as reasons not to treat the 2.1-billion-metric-ton projection as certain.

Permit requests across the United States already far exceed available grid capacity, and a sizable number of proposed buildouts face strong opposition. Those constraints can keep planned capacity out of the market, lowering both electricity demand and emissions in a buildout scenario.

The AI market remains active, but concern about hyperscaler economics could cause some hyperscalers and their projects to fail. That possibility adds another source of uncertainty to a forecast built around pending facilities rather than a guaranteed construction total.

For now, the 2.1 billion metric tons is a scenario estimate, not a commitment from data-center operators. By 2030, turbine prices and delivery slots, grid connections, permitting and hyperscaler economics will determine how close power-sector emissions come to the projection.