Author: Arnaud

For more than a decade, building a new coal-fired power plant in the United States looked politically, financially, and environmentally impossible. Coal was treated as a fuel of the past: too dirty, too controversial, and too exposed to regulation.

But the explosive growth of artificial intelligence and data centers may be changing that calculation.

The uncomfortable truth is simple: AI does not run on slogans. It runs on electricity. Vast, constant, reliable electricity. Every new hyperscale data center, every AI training cluster, every cloud-computing expansion adds another heavy load to a grid that is already under pressure from electrification, industrial reshoring, transmission constraints, and the retirement of older power plants.

This is why a new coal project in West Virginia, once almost unthinkable, is suddenly being discussed seriously.

The proposed TerraSpark Energy Campus would be a 1.6-GW greenfield coal plant using four 400-MW supercritical coal units. It would be paired from the start with carbon-capture technology, rather than treating emissions control as an afterthought. The project has received early-stage support from the U.S. Department of Energy for engineering, permitting, and technical studies.

That does not mean the plant will definitely be built. Far from it. The hard questions remain: total cost, financing, long-term customers, permitting, carbon-storage execution, and whether the carbon-capture technology can scale economically. But the fact that such a project is even moving through the development pipeline is significant.

It tells us something important about the new energy reality.

For years, policymakers assumed that coal could be retired while renewables, batteries, gas, nuclear, and transmission upgrades would fill the gap. In theory, that transition looked manageable. In practice, demand is now rising faster than many expected. AI has accelerated the problem. Data centers do not consume electricity like ordinary commercial buildings. They require huge amounts of power, often around the clock, and they demand reliability. A few hours of instability is not an inconvenience; it is a business risk.

That is where coal re-enters the discussion.

Coal’s great weakness is emissions. Its great strength is dispatchability. A coal plant can provide firm power day and night, regardless of weather, time of day, or battery duration. For grid planners and industrial users facing large, constant loads, that reliability still has value.

The new political language around coal is no longer simply “bring back the past.” It is being reframed as “firm power,” “energy security,” “industrial competitiveness,” and “baseload generation for AI and manufacturing.” Add carbon capture to the proposal, and developers can argue that modern coal may have a place in a lower-carbon grid, especially where power demand is rising faster than clean alternatives can be deployed.

Whether that argument succeeds is another matter.

The memory of failed or over-budget coal-with-capture projects still hangs over the sector. Investors will not fund nostalgia. They will want numbers. They will want contracted customers. They will want proof that carbon capture can work at scale, at acceptable cost, and without turning the plant into another expensive experiment.

But the bigger point is this: AI may be forcing the energy debate back toward realism.

A country cannot simultaneously demand rapid electrification, massive AI infrastructure, reshored manufacturing, domestic energy security, and the retirement of firm generation without replacing that capacity with something equally reliable. Wind and solar will remain important. Gas will likely continue to grow. Nuclear may become essential again. Batteries will help. But none of these options can be built instantly or everywhere.

That is why coal, once written off, may find a narrow but renewed opening.

Not as the old coal of the 20th century. Not as an unlimited comeback. But as part of a hard-nosed discussion about how to power an AI-driven economy that is consuming electricity at a scale few politicians seemed ready to confront.

The irony is striking. The technology sector, often presented as clean, modern, and futuristic, may end up creating demand for some of the oldest forms of firm power. AI may be digital, but its appetite is brutally physical: land, chips, cooling systems, transmission lines, transformers, and power plants.

The question is no longer whether coal is popular. It is whether the grid can meet the next wave of demand without keeping every dispatchable option on the table.

For years, a new coal plant in America sounded impossible.

Thanks to AI, it may now be a strong maybe.

Google’s Meitner Energy Center in the Texas Panhandle is being presented as a major clean-energy data center project, but its most important feature may be the inclusion of on-site gas-fired generation.

The co-located data center and power-generation complex will rely mainly on more than 1 GW of wind, solar, and battery storage.

However, Google and Intersect are also adding gas-fired generation to ensure reliable electricity when renewable output is not sufficient.

That gas component is significant. It shows that even large technology companies investing heavily in renewable energy still need firm, dispatchable power to support the continuous electricity demand of major data centers.

The decision also points to a clear economic advantage: access to nearby, relatively cheap natural gas. By combining renewables, batteries, and gas-fired generation, Google can reduce pressure on the local grid while securing a more reliable and cost-effective power supply for its data center operations.

The Meitner project therefore highlights a practical reality behind the clean-energy narrative: renewable power may provide much of the electricity, but traditional gas-fired generation remains essential for reliability.

The United States is no longer simply one of the world’s major oil producers. It is the world’s dominant producer. According to public data from the U.S. Energy Information Administration, the United States produced 21.91 million barrels per day of total oil in 2023. That was more than Saudi Arabia at 11.13 million barrels per day and Russia at 10.75 million barrels per day. Remarkably, on this broad “total oil” measure, U.S. production was even slightly higher than Saudi Arabia and Russia combined.

This matters because it changes the way energy security is discussed. For decades, global oil power was associated mainly with the Middle East and Russia. Today, the United States has become the central player in global liquid fuels supply.

The picture is slightly different when looking only at crude oil and lease condensate. On that narrower measure, the U.S. still leads each country individually, but not Saudi Arabia and Russia combined. In 2025, the U.S. produced around 13.58 million barrels per day of crude oil and condensate, ahead of Russia at 9.87 million barrels per day and Saudi Arabia at 9.51 million barrels per day.

The scale of the change is striking. U.S. crude oil production fell to around 5 million barrels per day in 2008, before the shale revolution transformed the sector. Horizontal drilling, hydraulic fracturing, and the rise of the Permian Basin helped push American output to successive records. By 2025, U.S. crude production had reached 13.6 million barrels per day, a new annual record.

This production strength is also relevant to the current debate around data centers, artificial intelligence, and electricity demand. Large data centers are often presented as clean-energy projects when they are connected to wind, solar, and batteries. But reliability still matters. When power must be available continuously, traditional energy sources remain central to the equation.

That is why projects such as Google’s Meitner Energy Center in the Texas Panhandle are important. They show the emerging model clearly: renewable generation and battery storage are used to reduce emissions and grid pressure, while on-site gas-fired generation provides firm, dispatchable power when needed.

The broader American energy picture helps explain why this model is attractive. The U.S. has become a huge producer of oil and natural gas, and regions such as Texas benefit from abundant nearby hydrocarbon resources. For large industrial power users, including data centers, access to reliable and relatively cheap gas remains a major advantage.

The conclusion is simple: the U.S. energy story is not only about renewables. It is also about scale, fossil fuel abundance, and reliability. America’s oil and gas production gives it a strategic advantage that Saudi Arabia and Russia no longer dominate in the same way.

Key source notes: EIA’s FAQ gives the 2023 total oil production ranking and explains that “oil” includes crude oil, other petroleum liquids, biofuels, and refinery processing gain. EIA also says U.S. crude oil plus condensate averaged 12.9 million b/d in 2023, broke the prior U.S. and global record, and that U.S. crude output recovered from a 2008 low of 5.0 million b/d. EIA reported U.S. crude oil production at a record 13.2 million b/d in 2024 and 13.6 million b/d in 2025. The 2025 international crude ranking shown in Graph 2 is based on Visual Capitalist’s EIA-derived Jan–Nov 2025 annualized data.