You’ve heard the panic. The U.S. grid is maxed out. AI data centers are guzzling electricity like there’s no tomorrow. Blackouts are coming. We need to build — fast — or the lights go out.
What if I told you the grid already has enough capacity to power hundreds of new power plants — and nobody’s using it?
Not in theory. Not in some lab. Right now. Today. Sitting there. Waiting.
The cheapest power plant in America is the one we already built but forgot to turn on.
Here’s the paradox that’s breaking the energy debate: everyone’s arguing about what to build next — more gas, more solar, more nuclear — but almost nobody is talking about the fact that the grid we have is being managed like it’s still 1985. The wires, the transformers, the substations — they’re carrying far less than they could, because the software deciding how to route power through them is dumb. Not slightly dumb. Spectacularly dumb.
We’re talking up to 300 gigawatts of latent capacity. To put that in perspective: that’s roughly the equivalent of hundreds of new power plants. Zero construction. Zero permits. Zero decade-long battles with NIMBYs. Just code.
Think about that for a second. Every time a politician proposes a new gas plant to “keep the lights on,” there’s a good chance that same capacity already exists — trapped inside a grid that doesn’t know how to use it efficiently. Every time an investor pours billions into generation assets, they might be building something that smarter software could render unnecessary.
We don’t have a generation problem. We have a distribution problem dressed up as a generation problem.
This is where it gets uncomfortable for a lot of people. The energy industry loves big assets. Concrete, steel, turbines — things you can touch, photograph at a ribbon-cutting, and depreciate over thirty years. Software doesn’t give you that. Software is invisible. Software doesn’t get a press conference with a hard hat and a golden shovel.
But software is the only thing that can move at the speed the grid now demands.
Here’s what’s actually happening on the grid right now: capacity is allocated conservatively, using static models that assume worst-case scenarios at every node, simultaneously, all the time. It’s like designing a highway where every lane is reserved for a crash that might happen. The result? Vast amounts of available capacity are locked away behind safety margins that modern, real-time optimization could dynamically manage.
Grid operators know this. Some of them are already doing it in pockets. But the prevailing narrative — the one driving policy, investment, and public anxiety — is still “build more.”
While everyone argues about whether to build gas or solar, the answer might be: build neither. Optimize what you have first.
Now, let me be clear about something. This isn’t a fairy tale. Software optimization isn’t going to replace every new power plant forever. Demand growth is real, especially with the AI buildout. Eventually, we’ll need new generation. But “eventually” is the key word. The question isn’t whether we need more capacity — it’s whether we need to spend a decade and hundreds of billions of dollars building it when 300 GW is sitting there, accessible, right now, for a fraction of the cost.
For grid operators, this should be a relief. For policymakers, this should be an urgent wake-up call. For investors betting on new fossil fuel peaker plants, this should be a cold sweat. And for consumers watching their electricity bills climb — this is the fight you should be paying attention to, because the cost of building new plants always ends up on your bill, one way or another.
The fossil fuel industry doesn’t want you to know this. The construction industry doesn’t want you to know this. Even some renewable developers don’t want you to know this, because “build more solar” is a simpler pitch than “optimize the grid’s existing capacity through software-driven load management and dynamic line rating.”
The most dangerous idea in energy isn’t a new technology. It’s the realization that we might not need one.
So here’s where I land. The next decade of energy policy shouldn’t start with the question “What do we build?” It should start with “What are we wasting?” The answer, it turns out, is roughly 300 gigawatts. That’s not a rounding error. That’s a redefinition of what grid expansion even means.
The power is already there. The question is whether we’re brave enough to use it — or whether we’d rather keep pouring concrete because it feels like progress.
FAQ
Q: If this is real, why isn't everyone doing it already?
A: Some grid operators are — in pockets. But the industry is built around physical assets, not software. Regulatory incentives reward capital expenditure, not optimization. You can depreciate a turbine; you can't depreciate a line of code. The institutional inertia is the real bottleneck, not the technology.
Q: Does this mean we never need to build new power plants again?
A: No. Demand growth from AI and electrification is real. But software optimization buys years of headroom at a fraction of the cost. The practical implication is simple: optimize first, build second. Right now, we're doing it backwards.
Q: Isn't this just a way to delay the renewable transition?
A: Actually, it accelerates it. Better grid utilization means existing renewables get curtailed less, fossil peaker plants run less, and the grid integrates variable generation more smoothly. The people who should be scared are the ones betting on new gas plants — not the ones building solar and wind.