The AI Energy Crisis: Why Google is Making a High-Stakes Bet on Carbon-Captured Natural Gas

It’s a pragmatic solution for 24/7 power. But is it truly “clean”?

Let’s be honest for a second. We all live in the “cloud.” Our emails, our photos, our late-night AI chats—they exist in this magical, intangible ether. But the cloud isn’t a cloud at all. It’s a network of massive, windowless, hyper-secure buildings called data centers, and they are hungry.

I’m talking ravenously, insatiably hungry for one thing: electricity.

And the new kid on the block, Generative AI, isn’t just hungry; it’s a competitive eater. The server-level computing power needed to train models like Gemini or power your ChatGPT queries is causing a spike in energy demand so massive it’s making utility companies sweat.

This puts companies like Google in a very tight spot.

For years, Google has been the poster child for corporate sustainability. They famously hit “100% renewable energy” back in 2017, meaning they bought enough solar and wind power annually to match their total global consumption. It was a huge win.

But it was a win on paper. An annual average.

At 3:00 AM in rainy Virginia, a data center is still churning, but the local solar panels are, well, asleep. At that moment, the data center is pulling power straight from the regional grid, which is very likely burning natural gas or coal.

This accounting-based “net-zero” wasn’t good enough for Google. In 2020, they set a “moonshot” goal that is infinitely harder: 24/7 Carbon-Free Energy by 2030.

This means every Google office and data center, in every hour of every day, must be powered directly by a clean energy source. No more averaging. No more wiggle room.

And that, my friends, is where the trouble starts. Solar and wind are fantastic, but they are “intermittent.” They don’t work when the sun isn’t shining or the wind isn’t blowing. Batteries help, but current technology can’t store grid-scale energy for days on end.

Google needs “clean, firm power”—a source that’s always on, 24/7/365, and doesn’t spew carbon. For a while, the only real options were geothermal and hydropower, but you can’t just build those anywhere.

So, facing the twin pressures of an exponential AI energy boom and an audacious 2030 deadline, Google just made one of the most pragmatic—and controversial—decisions in its history.

They’re investing in a natural gas power plant.

The “Clean” Gas Plant: Meet Broadwing Energy

Before you grab your pitchfork, this isn’t your grandpa’s gas plant.

Google just announced a landmark partnership with infrastructure investor I Squared Capital to develop Broadwing Energy, a brand-new, 400+ megawatt power plant in Decatur, Illinois.

The “magic” is this: the project is being built from the ground up to be fully integrated with Carbon Capture and Sequestration (CCS) technology.

Here’s the simple version of how it works:

1. The plant burns natural gas to generate electricity, creating CO2 emissions.
2. A chemical process “captures” that CO2 from the smokestack before it ever hits the atmosphere.
3. The captured CO2 is then compressed into a liquid-like state.
4. Finally, it’s injected over a mile deep into the earth into a permanent, EPA-approved geological storage well.

The plan for Broadwing is to capture and permanently bury over 90% of its carbon emissions.

This is the first-ever corporate agreement of its kind. Google is essentially acting as the anchor tenant, promising to buy most of the power this plant generates when it comes online around 2030. They are, in effect, creating a market for a technology that has historically struggled to get off the ground.

Why This Isn’t (Necessarily) a Greenwashing Scam

I know what you’re thinking. “Carbon capture? Isn’t that the boondoggle fossil fuel companies love?”

You’re not wrong to be skeptical. The history of CCS is littered with hideously expensive, over-budget projects that under-delivered and then quietly shut down.

But Google is arguing this time is different.

First, this isn’t “Enhanced Oil Recovery” (EOR). Most failed CCS projects were a thinly veiled excuse to get more oil. They’d capture CO2 only to pump it into aging oil fields to push more oil out, which completely defeats the purpose. The Broadwing project is explicitly for permanent sequestration. The carbon goes in the ground, and it stays there.

Second, this is “capture-committed,” not “capture-ready.” For decades, companies have built new gas plants with a vague promise to make them “capture-ready” someday… which almost never happens. This plant is being engineered with capture at its core.

Google’s argument is one of pragmatic engineering. They see this as a “bridge” technology. They believe that by underwriting this project, they can do for “clean, firm power” what they did for solar and wind a decade ago: prove it’s commercially viable, drive down costs, and create a blueprint for others to follow.

Their 24/7 data centers need firm power. Wind and solar provide intermittent power. This, they argue, is the only technology available right now that can fill the gap.

The “Hang On a Minute…” — The Big Problems with Clean Gas

Okay, let’s put our skeptic hats back on, because the problems are very, very real. This is not a silver bullet. It’s a deeply flawed and costly compromise.

1. The Methane Leakage Problem. This is the elephant in the room. Carbon capture grabs CO2 from the smokestack. It does nothing for the emissions from the rest of the supply chain.

The plant runs on natural gas, which is primarily methane. And methane, as a greenhouse gas, is over 80 times more potent than CO2 in the short term.

From the moment that gas is fracked out of the ground, through every leaky pipe, compressor station, and valve on its journey to Illinois, it is leaking methane into the atmosphere. CCS doesn’t touch this “upstream” pollution, which could be so significant that it negates the climate benefits of capturing the CO2 at the end.

2. The “Energy Penalty.” This is the part that feels like a cruel joke. The complex chemical and mechanical process of capturing, compressing, and injecting carbon takes a ton of energy to run.

Where does that energy come from? From the power plant itself.

It’s called the “energy penalty.” To produce the same 400 MW of net electricity and run the capture equipment, the plant will have to burn even more natural gas than a normal plant would. Estimates for CCS systems put this penalty anywhere from 15% to 25%. You have to burn more fossil fuel to clean up the fossil fuel. It’s a thermodynamic nightmare.

3. The Abysmal Track Record and Staggering Cost. Let’s not mince words: CCS has been a failure. The projects that have launched are few, far between, and have almost universally failed to meet their 90% capture targets. They are complex, they break down, and they are eye-wateringly expensive.

This project will produce some of the most expensive electricity on the grid. Google, with its $2 trillion market cap, can afford to pay a premium for this power as a “moonshot” R&D expense. But can the rest of the world? Does this create a scalable model, or does it just lock us into decades of mind-bogglingly expensive fossil fuel infrastructure?

4. The Moral Hazard. Finally, there’s the core argument from environmentalists: this is a dangerous distraction. Every billion dollars spent trying to “clean” a fossil fuel is a billion dollars not spent on developing cheaper, better batteries, next-generation geothermal, or other truly renewable 24/7 solutions.

Critics argue that CCS is an industry-backed mirage designed to do one thing: prolong the life of the fossil fuel industry for another 30 years under the guise of being “clean.”

So, What’s the Bottom Line?

This entire project is a massive, high-stakes, real-world experiment.

It’s a clash of two competing realities. Reality 1: The AI revolution we are all participating in is demanding an ungodly amount of 24/7, reliable power. Reality 2: Our only existing 24/7 power sources are fossil fuels (dirty) or nuclear (politically difficult).

Google, in a very Google-like move, is trying to solve this equation with brute-force engineering. They are betting that technology (CCS) can solve the problems created by their other technology (AI).

Is this a brilliant, pragmatic step toward a decarbonized future? A way to finally create the “firm, clean” power we’ve always needed to complement wind and solar?

Or is it a costly, methane-leaking compromise? A “green” fig leaf that allows the fossil fuel industry to survive and gives us permission to keep burning gas, all while distracting us from the real, harder work of building a truly renewable world?

The truth is, it’s probably a bit of both.

The Broadwing Energy plant in Illinois isn’t just a power plant. It’s a test case for the entire future of the energy grid. And our insatiable, 24/7 demand for data is footing the bill. The only question now is whether the result will be a climate breakthrough… or just the world’s most expensive compromise.