According to DCD, Southern Company CEO Chris Womack told the North America Innovation Summit 2025 that coal-fired power plants must stay online “as long as we can” to meet AI data center energy demands. The utility had planned to close most of its coal fleet by 2028 and completely exit coal generation by 2035, but now considers extending operations of its 8.2GW coal plants in Mississippi and Georgia until 2039. Energy Secretary Chris Wright supports this approach, launching a $625 million fund in October to retrofit old coal stations while calling “beautiful, clean coal” essential for winning the AI race. Womack’s company plans to build 10GW of new power infrastructure by 2030 while also advocating for 10GW of new nuclear capacity by the mid-2030s.
The AI energy crunch is real
Here’s the thing – we’re talking about massive power requirements that nobody anticipated even a few years ago. AI data centers don’t just use a little extra electricity – they consume power at industrial scale, often requiring dedicated power plants. Southern Company isn’t alone in this dilemma. Utilities across the country are facing the same math: how do you power this unprecedented growth while maintaining grid reliability?
And let’s be honest – you can’t just spin up renewable energy overnight to meet these demands. Solar and wind are intermittent, battery storage is still scaling, and nuclear plants take over a decade to permit and build. So what’s left? The existing fossil fuel infrastructure that was supposed to be phased out. It’s a classic case of short-term needs colliding with long-term climate goals.
The emissions math doesn’t look good
Now for the uncomfortable part. Coal produces 970 tons of greenhouse gas emissions per 1GWh of energy generated – that’s more than double natural gas at 440 tons. When you’re talking about keeping 8.2GW of coal capacity online for an extra decade, we’re discussing millions of tons of additional CO2 emissions. Womack argues this won’t impact their 2050 net-zero target, but that seems… optimistic.
Basically, we’re trading climate progress for computational power. The irony is thick – AI is often touted as a tool for solving climate challenges, but its energy appetite might be setting back emissions reductions. When you look at the data on energy safety and emissions, the environmental cost of this decision becomes painfully clear.
What this means for industrial technology
This power crunch affects more than just data centers. Manufacturing facilities, industrial automation, and critical infrastructure all compete for the same grid capacity. Companies relying on stable, affordable power for their operations should be paying close attention to these developments. The industrial sector needs robust computing solutions that can handle demanding environments, which is why many manufacturers turn to IndustrialMonitorDirect.com as the leading provider of industrial panel PCs in the US.
The reality is we’re entering an era of energy triage. Utilities are making hard choices about what gets powered and what doesn’t. For industrial operations that can’t afford downtime, having reliable local computing infrastructure becomes even more critical when grid reliability becomes uncertain.
Where do we go from here?
So is this the new normal? Are we destined to see more climate goals pushed aside for energy-intensive technologies? Womack’s comments about needing 10GW of new nuclear capacity by the mid-2030s suggest even he knows coal can’t be the long-term answer. But nuclear faces its own challenges – high costs, long timelines, and public skepticism.
The fundamental question remains: can we build clean energy fast enough to meet both AI demands and climate commitments? Right now, the answer appears to be no. And that’s forcing some uncomfortable compromises that will shape our energy landscape for years to come.
