Co-Located Data Centers: A Growing Debate
Dive into the co-location debate and discover key insights for managing data centers effectively in today's infrastructure landscape.
The argument started, as many infrastructure discussions do, with a power line.
When Crusoe Energy began pushing into the data center space — pivoting from its origins as a flared-gas computing company — it walked straight into a debate that Colorado regulators, utility planners, and grid operators hadn't fully worked out yet: What happens when a massive data center sits physically adjacent to a power generation source, and who gets to decide how that electricity flows?
Crusoe won't be the last developer to face this question. Co-located data centers — facilities built directly alongside or integrated with power generation assets, whether solar farms, natural gas plants, or battery storage projects — are multiplying fast, and the regulatory frameworks governing them are scrambling to keep up.
This isn't a niche technical dispute. It touches grid stability, ratepayer equity, permitting authority, and the basic question of who controls critical infrastructure. The debate is only getting louder.
What Co-Location Actually Means — and Why It's Complicated
At its simplest, a co-located data center shares a site or direct electrical interconnection with a generation asset. A hyperscaler builds next to a solar farm and draws power before it ever hits the grid. A crypto miner co-locates with a natural gas peaker plant and soaks up generation that would otherwise curtail. The proximity is intentional — it's a way to secure cheap, reliable, often low-carbon power without fighting for interconnection queue positions that can take five to seven years to clear.
The appeal is straightforward: skip the grid, skip the queue, and skip the interconnection costs that can run into the tens of millions of dollars per project.
But "skip the grid" is where the complications start. Utilities and grid operators aren't just bureaucratic gatekeepers — they're managing systems where every megawatt matters. When a 100 MW data center draws directly from a co-located solar project, that's 100 MW that may or may not show up in regional dispatch calculations. It creates blind spots. And blind spots in grid management create reliability risks that fall on everyone else connected to the system.
The definitional ambiguity compounds this. Is a co-located data center a "behind-the-meter" load? A microgrid? A separate entity that happens to share a fence line? Depending on how regulators classify it, the facility may or may not be subject to interconnection rules, standby charges, capacity obligations, or state public utility commission oversight. Different states are landing in very different places — and developers are acutely aware of which jurisdictions are friendlier than others.
The Competing Arguments — and What's Actually at Stake
Proponents of co-location make a compelling case. Data centers need power desperately — U.S. data center electricity consumption is projected to reach somewhere between 6% and 12% of national electricity demand by 2030, depending on the AI build-out trajectory you believe. Finding that power through new grid interconnections is slow and expensive. Co-location offers a path to scale faster, particularly for developers willing to build or partner with generation assets directly.
There's also an economic development argument. Co-located facilities create jobs, generate tax revenue, and in some configurations, actually accelerate the build-out of renewable generation that might not have been financeable on its own. A solar developer struggling to pencil out a merchant project in a low-price market suddenly has a creditworthy anchor tenant willing to sign a long-term power purchase agreement. That's not nothing.
The counterargument is sharper than it sounds: when large industrial loads exit the grid, the fixed costs of maintaining that grid don't disappear — they get redistributed to the customers who remain.
Utilities call this the "cost-shift" problem, and it's legitimate. Transmission lines, substations, and reserve capacity have to be maintained regardless of whether a given customer is drawing power through them. If data centers co-locate to avoid contributing to those costs, residential and commercial ratepayers pick up the slack. Some state commissions are starting to impose standby or exit fees specifically to address this. Others haven't touched it yet, leaving developers in a gray zone that's comfortable for now but unlikely to last.
There's also a grid reliability concern that deserves more attention than it typically gets. Data centers are not flexible loads by nature — they run continuously, at high utilization, with limited ability to curtail on short notice. When co-located with variable renewable generation, that creates a mismatch: the data center's demand is flat, but the solar or wind output isn't. Something has to bridge that gap, and if the facility isn't properly integrated with grid management systems, the answer might be "the grid absorbs it" — which is exactly the kind of unplanned burden grid operators are trying to avoid.
How Developers Are Navigating the Regulatory Maze
The smartest developers aren't waiting for regulators to sort this out — they're shaping the frameworks themselves. That means engaging early with state public utility commissions, participating in FERC proceedings on co-location policy (FERC has been actively examining this since 2023), and structuring projects in ways that reduce regulatory exposure.
Practically, this looks like a few specific strategies. Some developers are designing facilities with meaningful battery storage buffers, allowing the data center to absorb variability in generation output without pulling from the grid unexpectedly. Others are building in explicit demand response capabilities — contractual commitments to curtail load during grid stress events — as a way to demonstrate that co-located facilities can be grid assets rather than liabilities.
The developers who treat regulatory engagement as a cost center rather than a strategic function are going to find themselves on the wrong side of rulings they could have influenced.
Interconnection agreements for co-located projects are also getting more complex. Where a standalone data center might sign a simple retail service agreement, a co-located facility often needs a combination of agreements covering the generation asset's wholesale market participation, the data center's load obligations, and the treatment of any excess generation that flows onto the grid. Lawyers who specialize in energy regulatory work are seeing a meaningful uptick in exactly this kind of multi-layered structuring.
Where the Market Is Heading
The co-location model isn't going away — the economics are too attractive, and the hyperscalers driving demand are sophisticated enough to push it forward regardless of regulatory friction. What's changing is the sophistication of how these projects get structured.
Expect to see more integrated development: companies that control both the generation asset and the data center under a single corporate structure, reducing the contractual complexity and regulatory ambiguity that comes with arm's-length co-location deals. Nuclear co-location is also moving from theoretical to active — Microsoft's deal with Constellation at Three Mile Island being the high-profile example, but smaller advanced nuclear developers are in active conversations with data center operators about purpose-built co-location arrangements.
The battery storage component is likely to become non-negotiable. Not just because it solves the variability problem, but because regulators are increasingly using storage requirements as the mechanism for ensuring co-located facilities have some grid stabilization obligation. A project that comes in without a storage component is going to face harder questions than one that can demonstrate a meaningful buffer capacity.
Internationally, this is already a more mature conversation — data center operators in markets like the Netherlands and Singapore have been navigating co-location constraints for years, largely because grid capacity limitations forced the issue earlier. The U.S. market is about five years behind in regulatory sophistication on this specific question, which means there's a relatively short window to establish norms before the rules harden.
What Crusoe's Situation Actually Illustrates
Crusoe's position at the center of Colorado's co-location debate isn't just a company-specific story. It's a preview of how these disputes play out: a developer with a legitimate business model, a regulatory structure that wasn't built to accommodate it, and a set of stakeholders — utilities, ratepayers, grid operators, competing generators — who all have different interests and different amounts of political leverage.
The developers who navigate this well won't necessarily be the ones with the best technology or the cheapest power costs. They'll be the ones who understood early that infrastructure development in 2025 is as much a regulatory and political exercise as it is a construction and financing one — and who built their teams accordingly.
The debate over co-located data centers is, at its core, a debate about who bears the costs and captures the benefits of a major structural shift in how electricity gets consumed. That debate is far from settled. But the decisions being made right now — in state commissions, in FERC dockets, in boardrooms — will determine what the grid looks like for the next two decades.
Pay attention to who's in those rooms.
Explore more about the InfraSale Marketplace here.
INTERNAL LINK SUGGESTIONS
- [INTERNAL LINK: co-location benefits]
- [INTERNAL LINK: regulatory challenges]
- [INTERNAL LINK: energy market trends]