In parallel, Oracle is advancing Project Jupiter in southeastern New Mexico, a 2,500 MW campus running on methane fuel cells, entirely off-grid, with projected annual CO2 emissions of 10 million tons
Decision Focus
In May and June 2026, Denver, Jefferson County, and Longmont each moved to restrict new data center approvals, citing ratepayer cost exposure, water use, and emissions. Denver enacted a one-year moratorium. Jefferson County halted new zoning approvals. Longmont capped all new proposals at 100 MW. Colorado’s state legislature compounded the vacuum by letting two competing data center bills — one incentive-based, one mandate-based — die in its 2026 session, pushing the question to municipalities and the state PUC. For Global Heads of Data Center Energy, the local bans are a site selection signal, but the operational signal demanding attention is the off-grid model gaining scale in New Mexico and Utah, structured deliberately to sit outside any state clean energy mandate.
90-Second Brief
In recent days, colorado’s legislature failed to pass either a tax-incentive or clean-energy-mandate bill for data centers in 2026, leaving municipalities to fill the vacuum. Three jurisdictions moved within weeks to limit grid-connected development. In parallel, Oracle is advancing Project Jupiter in southeastern New Mexico, a 2,500 MW campus running on methane fuel cells, entirely off-grid, with projected annual CO2 emissions of 10 million tons. A separate Utah project proposes up to 9 GW of off-grid, natural-gas-powered capacity.
What Is Really Happening?
The Colorado situation is not primarily a land-use dispute. It is the visible surface of a regulatory arbitrage now structuring hyperscale site selection across the Mountain West.
Colorado’s clean energy targets — an 80% reduction in power sector emissions by 2030 — apply only to regulated utilities. A developer who builds a self-contained generation asset attached to a data center campus sits entirely outside that framework. That is not an ambiguous loophole; it is current law. As long as the project bypasses the regulated grid, Colorado’s environmental statutes have no mechanism to reach it.
Oracle’s Project Jupiter demonstrates the bypass at operational scale. The New Mexico campus is powered by methane fuel cells rather than conventional turbines — a technology Oracle positions as a partial emissions improvement — but its regulatory filings project 10 million tons of CO2 annually from that process alone. The campus is fully self-funded, deliberately isolated from the New Mexico grid, and located near natural gas pipeline infrastructure on land where water rights were acquired from local agriculture. New Mexico’s renewable energy standards have no reach over it.
Colorado currently lacks hyperscale activity at that scale. Its roughly 60 existing facilities are mostly small- to medium-sized, speculative builds leased to enterprise customers, drawing from the Xcel grid. Available natural gas, cheap land, and proximity to high-compute demand corridors make it an analytically plausible candidate for off-grid replication. Nothing in current Colorado law would prevent it.
Why It Matters for Global Heads of Data Center Energy
The moratorium wave directly affects site selection for grid-connected capacity in Denver and Jefferson County. Any operator with land under option or in early permitting faces a one-year pause, an uncertain 2027 legislative outcome, and no clear path to approval through normal zoning channels. Longmont’s 100 MW cap eliminates it as a hyperscale candidate regardless of what the state legislature does next year.
The more strategic pressure falls on the make-vs.-buy calculus. The off-grid ownership model removes several constraints that define conventional energy procurement: interconnection queue exposure, PPA counterparty risk, REC procurement costs, and utility rate increases passed through to co-located load. It replaces them with full ownership of fuel supply, emissions liability, and capital intensity. For operators who can absorb those tradeoffs, the model accelerates development timelines materially; for those who cannot, the competitive landscape in the Mountain West shifts.
For any team maintaining 24/7 carbon-free energy commitments, the off-grid gas model as deployed in New Mexico and Utah is not a minor deviation. A 10-million-ton annual CO2 position is not a rounding error in Scope 2 reporting or CDP disclosure. Operators benchmarking against Google, Microsoft, or Amazon’s clean energy posture face both a reporting exposure and a procurement positioning problem if the off-grid model becomes the regional norm.
The PUC proceeding Xcel has initiated in Colorado carries separate significance. If the commission determines that data centers must directly bear the incremental grid costs their load imposes — rather than socializing those costs across residential ratepayers — the economics of grid-connected development in the state shift. Environmental groups have successfully petitioned to participate in that proceeding, and the outcome could set a cost-allocation precedent with reach beyond Colorado.
Forward View
If the off-grid bypass continues to scale across the Southwest, regulatory pressure will follow the emissions rather than the grid connection. EPA jurisdiction and state air quality authority do not stop at utility boundaries. The New Mexico and Utah projects will attract that scrutiny as they approach full operation, and permitting friction and legal challenge exposure for large natural gas generation assets are material project risks that may not be fully reflected in current feasibility analyses.
Colorado’s 2027 legislative session is the next structural decision point. Whether legislators produce an incentive bill, a clean energy mandate, or another deadlock will determine whether Colorado formalizes a grid-connected development framework, opens an off-grid pathway with some environmental floor, or effectively cedes hyperscale activity to neighboring states. The Xcel PUC outcome, expected before that session, could harden or soften legislative positions depending on how the cost-allocation question resolves.
What Is Still Uncertain
The 2027 Colorado legislative outcome is genuinely open. Advocates for regulation say the 2026 bills provided a drafting foundation, but the lobbying dynamic that killed them has not changed. Whether the PUC proceeding produces a binding cost recovery framework — and on what timeline — is not determinable from current information.
Whether Oracle’s Project Jupiter and the Utah project reach full permitted capacity is also unresolved. Emissions volumes at that scale will attract legal challenge from environmental groups who acknowledged, as of mid-2026, that their tools to slow these projects are limited. The actual timeline from announcement to full operation, and the regulatory friction encountered along the way, should be tracked before treating either project as a confirmed precedent for the off-grid model at scale.
It is also not confirmed that any major operator is actively planning an off-grid, natural-gas-powered campus in Colorado. That scenario is analytically plausible based on current law and regional dynamics, but no such project has been announced.
One Question for Your Team
If a major operator replicates the off-grid natural gas model in Colorado before the 2027 legislature acts, what is your fallback procurement and interconnection position in the Mountain West — and can your current sustainability commitments survive that competitive and emissions landscape?
Sources
- Coloradosun — What’s the status of massive data centers in Colorado? Here’s what you need to know (Link)
